SummerFall 1999

Continues on Page 17Continues on Page 4

Beyond Arm Waving ThinkingCritically at Large ScalesLee Benda and Daniel MillerEarth Systems Institute Seattle

AbstractRecent advances in thinking about ecosystem behavior overthe past several decades have primarily taken the form ofqualitative concepts Although they represent major innova-tions in a number of fields they have also fueled a propensityfor arm waving and speculation when it comes to understand-ing how increasing scale changes our perceptions Further-more this limitation has also encouraged natural resourcemanagers and regulators to default to small scale reduction-ist approaches in their fields The present danger is thatscientific myopia in the watershed sciences focusing on smallscales is creating a number of problems that may weaken thesupport of policy makers for applying science to environmentalquestions and undermine the publicrsquos belief in the relevance ofscience in society To attack the problem of scale and hencecomplexity in watershed science management and regulationwill require developing new inter-disciplinary frameworks and theo-ries that address a broad range ofscales and scientific uncertaintyand that welcomes new forms ofknowledge In this article we out-line several universal principlesthat can be used to understand howmeasurement scale influences ourperception of environments Whenlinked with an understanding oflandscape processes they allow usto predict how unique mixtures ofclimate topography vegetationchannel networks and basin scaleimpose first-order constraints onspatial and temporal patterns ofenvironmental variability at anyspatial scale This form of knowledge isembodied in probability and frequency distributions and theyreveal the consequences of multiple interactions over mul-tiple scales within landscapes We present brief examples ofhow the behavior of forest ages landsliding sedimentationand large organic debris loading to streams varies according tocombinations of space and time scales Finally we suggesthow this type of understanding might play out in watershedscience resource management and regulation

Timber Harvest and SedimentLoads in Nine NorthernCalifornia Watersheds Based onRecent Total Maximum DailyLoad (TMDL) StudiesSharon H Kramer Martin Trso and NoahHumeStillwater Sciences Arcata

IntroductionThis review provides an assessment of sediment sourceanalyses from recently completed Total Maximum DailyLoad allocations (TMDLs) in nine northern Californiawatersheds in the range of the coho salmon (Oncorhynchuskisutch) The goal of this review is to address sedimentyields associated with forest management activitiesparticularly after the promulgation of forest practiceregulations Each TMDL provides a sediment sourceanalysis based upon different time frames and sourcecategories However this review does not analyze the

appropriateness of the methodol-ogy used or the results obtainedfor sediment source assessmentsin individual TMDLs A moreextensive analysis of the back-ground data used to prepare theTMDLs would be required toverify sediment source categoriesand to separate out the potentialeffects of large storm events thatvary substantially in magnitudeand frequency Using the pub-lished data available fromTMDLs we present a discussionof mechanisms of sedimentdelivery sediment source analysismethodology and a comparison of

sediment loadings due to timber harvest as well as sedi-ment production due to other associated forest managementactivities and natural sources

Causal Mechanisms for Sediment Delivery to streamsIn steep dissected and soil-mantled hillslopes of the humidand forested Pacific Northwest region mass wastingprocesses (ie landslides creep and biogenic transport) arenaturally a dominant erosion process on hillslopes and

Summer 2001Volume 10 No 1

2 WMC Networker Summer 2001

The Watershed ContinuumMoving Toward ConsilienceA cross-disciplinary group is developing a multimedialearning system and courseware development tool tofacilitate the discovery of interdisciplinary concepts novelsyntheses and consilience

A group of scientists and educators are striving to build an information container thatwill enable consilience the ldquojumping together of knowledgerdquo as recently described andelucidated by EO Wilson in his book Consilience The Unity of Knowledge

Working from Corvallis Oregon and Arcata California the group is building a

WATERSHED

MANAGEMENT

COUNCILNETWORKERis a publication of the

Watershed Management Councilco PSRP

University of California at DavisOne Shields AveDavis CA 95616

httpwwwwatershedorg

Rick Kattelman President

Chuck Slaughter President-Elect

Jim Bergman Secretary

Terry Kaplan-Henry Treasurer

MEMBERS-AT-LARGE

Robert CoatsJohn CobournPete HawkinsPolly HaysGeorge IceDonna MeyersDennis Pendleton

NEWSLETTER AND WEBSITE

Michael Furniss Editor and Webmaster

Kathryn Ronnenberg Production

MEETING DATESThe WMC Board of Directors meetsquarterly All WMC members arewelcome to attendContact a boardmember to arrange to attend a meetingor discuss any ideas or issues for theCouncil

MEMBERSHIPDues are $20 per year Please use themembership application form on theback page of this issue to join Forinquiries for subscription problems call(510) 273-9066 (WMCrsquos voice mail) orsend an email tosubscriptionswatershedorg

SUBMISSIONS WELCOMEThe WMC Networker welcomes allsubmissions All copyrights remain withthe authors Email or disk versions areappreciated Please keep formatting to aminimum Mail submissions to PresidentRick Kattelmann at

rickwatershedorg

Incoming Presidentrsquos ColumnFor this first column from the WMCrsquos incoming president for 2001-2002 there are

two critical items of business to encourage every readerrsquos participation in our organi-zation and to thank Sari Sommarstrom and Mike Furniss for their exceptional serviceto the Watershed Management Council

First the encouragement the Watershed Management Council is a member-basedorganization The WMC acts for the benefit of its members and as a vehicle for itsmembers We formed this outfit back in 1986 to pool our resources in advancing thefield of watershed management The charter members felt that a ldquoWatershed Man-agement Councilrdquo could do more than most of us can as individuals However asindividuals we need to contribute to the group by sharing our experience and ideaswith the watershed management community The WMC is primarily an educationalorganization We try to educate our membership our colleagues and associates policymakers and the public at large about the benefits potentials tools and techniquesof watershed management So far wersquove tried to accomplish that mission through theNetworker our web site our conferences and field trips our policy forums and co-sponsorship of events organized by other groups All these activities (and others wehavenrsquot yet explored) require help from the individual members We need articles forthe Networker we need material for the website we need speakers for conferencesfield trips and forums we need organizers for events and we need fresh ideas for howthe Watershed Management Council can serve its members

Second the thanks two of the individual members who have given tremendous helpand service to the Watershed Management Council are stepping into less-active roleswith the WMC this year Sari Sommarstrom is now Past-President after a verysuccessful two-year term Sari has been with the WMC since its beginning was anactive board member contributed many articles to the Networker and conferenceproceedings helped organize at least three WMC conferences was an outstandingPresident and organized the four California Watershed Forums in 1999 and 2000The Watershed Management Council is most grateful for all of Sarirsquos efforts with theorganization and her service to the field of watershed management Sari will berepresenting the WMC at the National Watershed Forum to be held this summer

Thanks part 2 Mike Furniss is leaving the editorship of the WMC Networker aftereleven years Obviously we canrsquot thank him enough for creating an excellent publica-tion and continually improving its quality over the years The high frequency that I seecitations to WMC Networker articles in research papers state-of-the-art summariesand technical reports is testimony to the quality of our Networker under Mikersquosleadership and fine-tuning Wersquove also heard plenty of compliments about the read-ability and visual appearance of the Networker Again Mike has been responsible forthe excellent layout and ldquolookrdquo of the publication All the past issues of the Networkersince 1990 are available on the WMC website (wwwwatershedorg) which Mike hasgraciously consented to keep maintaining So Irsquod like to extend our deepest apprecia-tion to Mike Furniss for being a fantastic editor and webmaster

Finally we are now in need of a Newsletter Editor Do you have an interest Do youknow someone who might You donrsquot need to know anything about layout or pagedesign publishing software or print shops All that can be contracted for This is agreat opportunity for someone Mike Furniss told me recently that his entire careerchanged for the better after he became Editor Call me if you would like to discussthis-

Continues on next page column 2

2 WMC Networker Summer 2001

The Watershed ContinuumMoving Toward ConsilienceA cross-disciplinary group is developing a multimedialearning system and courseware development tool tofacilitate the discovery of interdisciplinary concepts novelsyntheses and consilience

A group of scientists and educators are striving to build an information container thatwill enable consilience the ldquojumping together of knowledgerdquo as recently described andelucidated by EO Wilson in his book Consilience The Unity of Knowledge

Working from Corvallis Oregon and Arcata California the group is building a

WATERSHED

MANAGEMENT

COUNCILNETWORKERis a publication of the

Watershed Management Councilco PSRP

University of California at DavisOne Shields AveDavis CA 95616

httpwwwwatershedorg

Rick Kattelman President

Chuck Slaughter President-Elect

Jim Bergman Secretary

Terry Kaplan-Henry Treasurer

MEMBERS-AT-LARGE

Robert CoatsJohn CobournPete HawkinsPolly HaysGeorge IceDonna MeyersDennis Pendleton

NEWSLETTER AND WEBSITE

Michael Furniss Editor and Webmaster

Kathryn Ronnenberg Production

MEETING DATESThe WMC Board of Directors meetsquarterly All WMC members arewelcome to attendContact a boardmember to arrange to attend a meetingor discuss any ideas or issues for theCouncil

MEMBERSHIPDues are $20 per year Please use themembership application form on theback page of this issue to join Forinquiries for subscription problems call(510) 273-9066 (WMCrsquos voice mail) orsend an email tosubscriptionswatershedorg

SUBMISSIONS WELCOMEThe WMC Networker welcomes allsubmissions All copyrights remain withthe authors Email or disk versions areappreciated Please keep formatting to aminimum Mail submissions to PresidentRick Kattelmann at

rickwatershedorg

Incoming Presidentrsquos ColumnFor this first column from the WMCrsquos incoming president for 2001-2002 there are

two critical items of business to encourage every readerrsquos participation in our organi-zation and to thank Sari Sommarstrom and Mike Furniss for their exceptional serviceto the Watershed Management Council

First the encouragement the Watershed Management Council is a member-basedorganization The WMC acts for the benefit of its members and as a vehicle for itsmembers We formed this outfit back in 1986 to pool our resources in advancing thefield of watershed management The charter members felt that a ldquoWatershed Man-agement Councilrdquo could do more than most of us can as individuals However asindividuals we need to contribute to the group by sharing our experience and ideaswith the watershed management community The WMC is primarily an educationalorganization We try to educate our membership our colleagues and associates policymakers and the public at large about the benefits potentials tools and techniquesof watershed management So far wersquove tried to accomplish that mission through theNetworker our web site our conferences and field trips our policy forums and co-sponsorship of events organized by other groups All these activities (and others wehavenrsquot yet explored) require help from the individual members We need articles forthe Networker we need material for the website we need speakers for conferencesfield trips and forums we need organizers for events and we need fresh ideas for howthe Watershed Management Council can serve its members

Second the thanks two of the individual members who have given tremendous helpand service to the Watershed Management Council are stepping into less-active roleswith the WMC this year Sari Sommarstrom is now Past-President after a verysuccessful two-year term Sari has been with the WMC since its beginning was anactive board member contributed many articles to the Networker and conferenceproceedings helped organize at least three WMC conferences was an outstandingPresident and organized the four California Watershed Forums in 1999 and 2000The Watershed Management Council is most grateful for all of Sarirsquos efforts with theorganization and her service to the field of watershed management Sari will berepresenting the WMC at the National Watershed Forum to be held this summer

Thanks part 2 Mike Furniss is leaving the editorship of the WMC Networker aftereleven years Obviously we canrsquot thank him enough for creating an excellent publica-tion and continually improving its quality over the years The high frequency that I seecitations to WMC Networker articles in research papers state-of-the-art summariesand technical reports is testimony to the quality of our Networker under Mikersquosleadership and fine-tuning Wersquove also heard plenty of compliments about the read-ability and visual appearance of the Networker Again Mike has been responsible forthe excellent layout and ldquolookrdquo of the publication All the past issues of the Networkersince 1990 are available on the WMC website (wwwwatershedorg) which Mike hasgraciously consented to keep maintaining So Irsquod like to extend our deepest apprecia-tion to Mike Furniss for being a fantastic editor and webmaster

Finally we are now in need of a Newsletter Editor Do you have an interest Do youknow someone who might You donrsquot need to know anything about layout or pagedesign publishing software or print shops All that can be contracted for This is agreat opportunity for someone Mike Furniss told me recently that his entire careerchanged for the better after he became Editor Call me if you would like to discussthis-

Continues on next page column 2

WMC Networker Summer 2001 3

product that will help students of all ages perceive connectionsbetween myriad knowledge resources that exist and are nowmore accessible than ever with electronic communications toolsThe team is designing and building a ldquozoomingrdquo user interfacea body of compelling scientific content organized around aconceptual taxonomy The Watershed Continuum will containand deliver courseware that is readily shareable extensiblefriendly and accessible An abbreviated version of the cur-rent project description follows

Guiding principles of the Continuum ProjectThe Continuum Project is about merging experience and

practice with immediate access to the wealth of informationexpertise and collegiality that has suddenly become availablevia computing and communications It is about realizing thepotential in information technology to achieve a higher level ofunderstanding to bridge disciplines once separated by lan-guage and proximity to link ideas and leverage the ability ofscience to paint a more vivid picture of earthrsquos processes

The Continuum Project is no more or less about science thanit is about philosophy and learning The key success factor intodayrsquos information-rich societymdashopennessmdashreflects the long-understood truth that collaboration fosters scientific under-standing Our government and social institutions are strongerwhen there is active participation by an informed citizenryThe Internet is upon us because of open standards forinteroperation The processes of nature are open for all to seeif they will observe In that spirit this project embraces openstandards for information exchange and contains resourcesgleaned from the public domain or generously donated byauthors and speakers It hopes to foster sharing and dialogacross disciplines so each can learn from and strengthen theothers and move several steps closer to an integrated under-standing of how the world works

While the Continuum Projectrsquos initial focus is to provide a

habitat and the watershed as a whole A watershed assess-ment and action plan were completed and adopted in 1995with several subsequent revisions The Associationrsquos list ofaccomplishments includes a great variety of projects such as80 miles of riparian livestock exclusion fencing and tree plant-ing replacement andor repair of 22 culverts to improve fishpassage and enhancement or creation of 12 off-channel wet-lands More than 200 private landowners within the CoquilleRiver basin have completed such restoration projects and 25more are on a waiting list for future cooperative projects Thegroup was recognized previously for its outstanding watershedwork by the US Forest Service Bureau of Land Managementand the Oregon Private Industry Council The Coquille Water-shed Association is to be congratulated for its success ininvolving a diverse array of interests in supporting and imple-menting watershed management goals

The Awards Program will be an ongoing activity of the Water-shed Management Council We are seeking a financial sponsorfor the awards Any suggestions in this regard would beappreciated Members should be thinking about potentialnominees for the next pair of awards in 2002 We hope that theWMC awards program will provide greater public visibility forindividuals and groups actively ldquoadvancing the art and scienceof watershed managementrdquo-

The first recipients of the WMCrsquos new leadership awards wereannounced at our 8th Biennial Conference at Asilomar LouiseSolliday most recently Watershed Advisor to the Governor ofOregon received the 2000 John Wesley Powell Award for ldquoAnIndividual Making an Outstanding Contribution to Water-shed Management in the Western United Statesrdquo The Co-quille Watershed Association of Coquille Oregon received the2000 Walter C Loudermilk Award for ldquoAn Organization Imple-menting an Outstanding Watershed Restoration Programrdquo

WMC President Sari Sommarstrom presented the awardswhich included a personally engraved plaque and anautographed copy of Dr Luna Leopoldrsquos renowned book A Viewof the River While an award program was suggested in theCouncilrsquos 1986 by-laws it took Board Member Robert Coats toofficially initiate these two awards last summer The CouncilBoard is pleased to finally have in place this program for morerecognition of individuals and organizations contributing tothe field of watershed management in the western states

Among the nomination letters we received for Louise Sollidaywas one from Governor John Kitzhaber We wish to quote a fewexcerpts from that letter

ldquoIt is with great pleasure that I recommend for the JohnWesley Powell Award an individual that has been an out-standing leader in watershed management in Oregon Whilethousands of dedicated and talented individuals contribute towatershed restoration and all of them deserve awards Iwould argue that Louise Solliday has provided the essentialstrategic leadership that has allowed local leadership to flour-ishrdquo

ldquoOver the last four years Ms Solliday has made it her toppriority to identify the strategic needs of local watershedrestoration efforts and has worked tirelessly to fill thoseneeds She recognized that the many volunteers involved inrestoring their watersheds need simple tools and lsquohow torsquomanuals to apply to their watersheds Ms Solliday organizedthe the development of restoration guidelines and a watershedassessment manual and worked with the federal agencies toapprove these guidelines Ms Solliday recognized that if wedid not make watershed restoration relatively easy it wouldnot succeed Toward this end she has worked hard to coordi-nate and streamline federal and state rules governing streamrestorationrdquo

ldquoFor three years Ms Solliday chaired the grant program thatOregon established to fund local watershed restoration ef-forts She guided the Watershed Enhancement Board througha transition from a small granting entity to an organizationthat provides more than $30 million a biennium toward water-shed restorationrdquo

ldquoAgain while I believe that thousands of Oregonians aredeserving of the John Wesley Powell Award I am a strongsupporter of Ms Louise Sollidayrdquo

Shortly before the WMC conference Ms Solliday was pro-moted to be the Governorrsquos Assistant for Natural Resources

WMCrsquos first Walter C Loudermilk Award for an organizationwas presented to Paul Heikkilla former Chair on behalf of theCoquille Watershed Association The group was formed in1993 by watershed residents who were concerned about aquatic

Continues on Page 33

The Continuum Continued

Louise Solliday and Coquille Watershed Association Honored inFirst WMC Awards

4 WMC Networker Summer 2001

IntroductionMany environmental issues in the watershed sciences involv-ing forestry geology hydrology fisheries etc depend on inter-actions of numerous landscape processes acting over a largerange of spatial and temporal scales Consequently scientistsand managers have been seeking to understand the role ofscale in their respective disciplines for more than two decades(de Boer 1986 Allen and Starr 1982 Folt et al 1998) Forexample recognizing hierarchical time and space domains insurface processes brought issues of scale to the forefront in thescience of community and landscape ecology (Wiens 1984Frissel et al 1986) Motivated in part by ecologists watershedscientists also have recognized the need to transfer under-standing across spatial and temporal scales in the study oflandscapes (Swanson et al 1988 Benda et al 1998)

Despite this attention our understanding of how increasingthe dimensional scales of perception would play out in water-shed science management and regulation is limited Forexample it has proven difficult to extend channel reach-scalemeasurement and understanding obtained over a period of afew years to larger spatial and temporal scales (ie segmentto network over decades to centuries) because of un-reconciledstochastic processes and the lack of theory that would allowbridging between data gaps (Benda et al 1998) As a conse-quence most resource managers focus their planning activi-ties at small scales (harvest unit single rotation) with deci-sion making therefore hinged primarily on market forces andchanging environmental regulations Furthermore althoughregulators recognize the need for understanding resource man-agement and impact assessment at large scales they defaultto small scale reductionist approaches primarily because ofthe lack of leadership in this area by the scientific communityNevertheless this problem is on the radar screen as evidencedby a growing consensus calling for incorporating dynamics ordisturbance (ie increasing scale) into natural resource sci-ence and management (Botkin 1990 Naiman et al 1992Reeves et al 1995 Reid 1998 Dunne 1998 Lackey 1998Benda et al 1998 Maurer 1999) In addition adapting scienceto questions involving larger space and time scales is becom-ing a central tenet in new and creative forms of thinkingembodied in systems analysis (Weinberg 1975 Allen andStarr 1982) macro ecology (Mauer 1999) and complexityscience (Kellert 1993 Waldrop 1992 McIntrye 1998)

The purpose of this article is to outline highlight severaluniversal principles that address the effects of increasingscale in watersheds or landscapes Our aim is to equip readerswith a handful of tools that should allow them to think criti-cally regarding the role of scale in watershed science manage-ment and regulation in their home landscapes in the hope ofmoving beyond the hand waving stage In addition we empha-size how large-scale attributes of landscapes (ie mixtures ofclimate topography vegetation network geometry and basinscale) impose major constraints on the space and time struc-ture of environmental variability (ie the disturbance re-gime) embodied in forests erosion large organic debrisstreamflow channel conditions and aquatic habitats Be-cause the study of scale in the watershed sciences is a rela-tively new and rapidly evolving field we admit that under-

standing how increasing scale might play out in natural re-source management and regulation (forestry fisheries etc) isat an incubation stage We leave that topic up to the unbridledimagination of the reader

Universal Principles of ScaleA Landscape Parameter of Frequency and ProbabilityDistributions

Simply the system behavior of a landscape can be describedas the temporally changing values of an attribute at a singlelocation (forest ages sedimentation large woody debris etc)or the variation in these attributes over many locations in awatershed in a single year (Benda et al 1998) Hence thesystem behavior of landscapes cannot be understood solelywith single value parameters Landscape parameters mustbe multiple valued to reflect temporal variability and spatialheterogeneity key aspects to understanding geomorphic and

0

1

2

3

0 200 400 600 800 1000Years

Dep

th (

m) Temporal Sample

0

10

20

30

40

0 1 2 3Aggradation Depth (m)

Y

ears 100 yr

1000 yr

Spatial Sample

0

1

2

0 2 4 6 8 10

Distance Along Channel (km)

Dep

th (

m)

0

10

20

30

40

0 1 2Depth (m)

L

engt

h

1 km

10 km

Beyond Arm Waving Thinking Critically at Large ScalesContinued from Page 1

Figure 1 In temporally variable environments the longer thetime of observation the greater the probability of seeing alarge event here represented with a simulated time series ofsediment depth at a channel cross section Aggradationalevents create terraces with the highest terrace recording therarest events The frequency of aggradation influences thevegetation communities inhabiting each terrace

Figure 2 In spatially variable environments the larger thearea observed the greater is the probable range of observationsmade illustrated here with simulated sediment depths along a10-km channel reach

WMC Networker Summer 2001 5

ecologic behavior A distribution of values inexorably ariseswhen attributes are viewed either over long periods or largeregions of space it cannot be inferred from a point observationIn the parlance of the evolving field of complexity sciencefrequency or probability distributions comprise an emergentproperty that reveals the consequences of multiple interac-tions over multiple scales within landscapes

For example soil depth is a parameter and is used in manyapplications such as landslide prediction groundwater mod-eling etc The distribution of soil depth is however a param-eter itself because it contains information about the range andfrequency of values Hence when seeking to understand therole of scale distributions of parameters (such as soil depthforest ages rainfall etc) will be integrated into quantitativerelationships that predict watershed (system) behavior in theform of other distributions (Benda et al 1998) For examplethe probability distribution of climate coupled with frequencydistributions of topographic and vegetative attributes willyield a probability distribution of erosion and sediment flux(Benda and Dunne 1997) Distributions because they mea-sure temporal variability and spatial heterogeneity are sen-sitive to changing scales and they also indicate how measure-ment scale influences our perception of environments

Relationships Among Distributions Time and Space

Most physical and biological landscape processes are stronglydependent on scale For example the occurrence of distur-bances such as earthquakes landslides fires storms floodsetc is dependent on time scale These dynamic surface pro-cesses contributes to habitat diversity a multi-valued at-tribute that depends on spatial scale Hence intrinsic rela-tionships between temporal and spatial distributions andscale comprise a fundamental element for understanding howphysical and biological processes are interrelated Five uni-versal scaling relationships are briefly outlined below someare trivial and this section is meant as a review for manyreaders

Scaling in time (disturbance regime at a point) Manystochastic agents including earthquakes storms fires floodsand erosion act to create and modify channel and riparianlandforms The time required to observe the full range ofevents responsible for the full suite of riverine morphologies(eg disturbance regime or range of variability) depends on theoperative processes In general infrequent but large magni-tude events are responsible for large morphologic changes with

long legacies processes located in tails of probability distribu-tions (Dunne 1998) The structure of that temporal variabilityis increasingly apparent over larger time scales (Figure 1)

Scaling in space (many points diversity at a single time)An observer (human or otherwise) following a stream channelencounters a diversity of attributes (eg particle sizes pooldepths etc) that originate from temporal variability (Figure1) and deterministic topographic heterogeneity (ie system-atic decreases in gradient downstream for example) Therange of that diversity increases with the distance traversed(Figure 2) Thus the range of habitat types available to anorganism increases with increasing scale of movement Aquaticorganisms may have evolved different life history strategies in

10-kmChannel Network

1 km

02

04

06

08

0 25 50 75 100Year

Mea

nD

epth

(m)

0

10

20

30

0 05 1Depth (m)

L

engt

h

10 km

02

04

06

08

0 25 50 75 100Year

Mea

nD

epth

(m)

yr 40yr 45

0

10

20

0 05 1Depth (m)

L

engt

h yr 40

yr 45

1-km

reach

(A)

(B)

(A)

(B)

Sediment Depth

Y

ears

Sediment Depth

Y

ears

Sediment Depth

Y

earsChannel

Network

Sediment Source Area

Figure 4 Integration of multiple inputs via routing through achannel network causes the distribution of sediment flux andassociated storage to become more symmetrical with lowervariance downstream

Figure 3 In environments that vary in space and time thegreater the number of elements included in a population theless variable the distribution over time Here mean sedimentdepth over a 1-km length and a 10-km length of channel iscompared over a 100-yr time span Although the range ofvalues sampled over the 10-km length is greater the differenceis the distribution of values and in the mean value from yearto year is less

6 WMC Networker Summer 2001

response to the spatial and temporal scales of habitat hetero-geneity A similar pattern would be evident with landscapeattributes not linked to a channel network such as forestscomprised of different age stands or ages of landslide scars

Scaling in space and time (many points over time habi-tat stability) Stochastic agents of disturbance in combina-tion with time-invariant topographic features (ie channelgradients valley width etc) will change the set of attribute

values found at any single location over time (Figure 1) How-ever temporal variability can also be viewed at larger spatialscales say a population of hillslopes or channel reaches forexample The third universal relationship between distribu-tions and scale indicates that the extent a population ofsomething changes (ie for example channel reaches in termsof the shape of the distribution or of some statistical momentsuch as the mean mode or variance) is dependent on the size

150 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

500 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

200 300 400 500

150 years

0

2

4

6

8

10

12

Magnitude

Y

ea

rs

0 100 200 300 400 500

500 years

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

50 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

Time Series

0

100

200

300

400

500

600

700

800

0 100 200 300 400 500

Year

Mag

nitu

de

50 years

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

FIE

LD

ME

ASU

RE

ME

NT

S

MO

DE

L PR

ED

ICT

ION

Figure 5 At sufficiently large space and time scales the frequency distribution of a spatial ensemble of elements (Figure 2) maybecome statistically similar to a temporal distribution of that attribute at a single location (Figure 1) Similarity between largespatial samples and large temporal samples is referred to as an ergodic relationship (Hann 1977) and it may be used to infertemporal behavior from a spatial sample Although complete statistical equivalence is unlikely location-for-time substitution(Brundsen and Thornes 1978) allows for an understanding of temporal change from only a spatial sample

WMC Networker Summer 2001 7

of the population of elements Holding the mean size of adisturbance such as a fire or landslide constant the probabil-ity that the distribution of values will change over any giventime increases as the size of the population decreases (Figure3) This scaling relation is also observed in the temporalvariability of the mean value of the attribute under consider-ation This is simply a consequence that something (firelandslide flood) does not happen at the same intensity (oreven occur) every place at once The statistical stability of aspatial distribution of physical habitats (ie resilience toshifts over time) may influence such things as population sizespatial distribution and dispersal strategies

Scaling in time modulated by channel networks Scalingrelationships one through three are not only applicable tolandscapes or channel networks they are applicable to anyenvironment including galaxies or grains of sand on a beachThe fourth universal scaling relationship applies particularlyto branching networks such as a channel system Over longtime periods a channel network samples from a multitude ofindividual sources of mass flux (such as sediment and LWD)each with their own probability distribution (Figure 1) Theintegration of all sampled sources yields a long-term distribu-tion of mass flux through the network that evolves down-stream into more symmetrical forms a demonstration of thecentral limit theorem (Benda and Dunne 1997 Figure 4) Thisspatial evolution of the probability of channel and ripariandisturbances may impose spatial organization on aquatic andriparian habitats and on the species that inhabit them

Similarity Between Space and Time Variability (usingfield clues) At sufficiently large space and time scales thefrequency distribution of a spatial ensemble of elements (Fig-ure 2) may become statistically similar to a temporal distribu-tion of that attribute at a single location (Figure 1) thispotential commensurability is shown in Figure 5 For statis-tical similarity to hold all forcing functions must be tempo-rally stationary (over certain intervals) Similarity betweenlarge spatial samples and large temporal samples is referredto as an ergodic relationship (Hann 1977) and it may be usedto infer temporal behavior from a spatial sample Althoughcomplete statistical equivalence is unlikely location-for-timesubstitution (Brundsen and Thornes 1979) allows for an un-derstanding of temporal change from only a spatial sampleThis technique is particular useful for evaluating landscapebehavior including model predictions of same that encompassdecades to centuries during short human life spans Next weexamine some of the ramifications of increasing spatial andtemporal scale on watershed science management and regu-lation

Increasing Scale Implications for WatershedScienceCircumventing Limitations in Understanding CoarseGraining

Landscape interactions create patterns in the temporal be-havior and spatial distribution of watershed attributes thatmay be discernible only at large temporal and spatial scales(Figures 1 - 4) Because of the absence of long-term datacomputational models are needed to identify these patternsand the interactions that created them

Ideally large-scale computational models would be built uponphysics-based understanding of all relevant interdisciplinaryprocesses However there is a growing consensus that at leastin the near future there will be theoretical and technicalimpediments to developing large-scale environmental modelsbased on smaller-scale processes (Weinberg 1975 Dooge 1986)To circumvent present limitations watershed-scale modelscould be constructed whereby some small-scale processes areignored or are subsumed within larger-scale representationsof those processes a strategy commonly applied by physicistsand referred to as coarse graining (Gell-Mann 1994) In prac-tice in the watershed sciences this often requires combiningby means of mathematical synthesis and computer simula-tion empirical knowledge with theoretical reasoning avail-able at smaller scales to produce new understanding at largerscales (Roth et al 1989 Benda and Dunne 1997) The objec-tive of this approach would not be precise predictions aboutfuture states at individual sites but rather new testablehypotheses on large-scale interactions of climate topographyvegetation and riverine environments This approach has ahistory in the study of certain hydrological problems at largescales (Smith and Bretherton 1972 Rodriguez-Iturbe andValdez 1979)

Coarse-grained modeling is well suited to the use of distribu-tions and exploring the effects of scale on them (ie Figures 1ndash 5) System models by definition produce synthetic timeseries of watershed behavior (ie forest fires storms erosionchannel changes etc) In evaluating model predictions overhuman time frames (ie for hypothesis testing or environmen-tal problem solving) the typical absence of long-term dataforces us to consider a population of watershed elementssampled over a large spatial area Here again point observa-tions are insufficient to reveal system behavior and the distri-bution parameter in numerical models allow us to link behav-ior over large time and space scales to field observations madeover short times but large areas

Although the coarse-grained approach is taken by necessity inthe study of complex environmental systems there is consid-erable potential for obtaining new insights and understand-ing Focusing on small-scale processes often precludes seeingthe ldquobig picturerdquo that is the emergence of patterns and pro-cesses unseen at small scales Indeed the description of large-scale patterns of behavior even in the absence of mechanisticunderstanding for all of the observed processes is a hallmarkof the study of complex systems (Waldrop 1992 Kellert 1993Gell-Mann 1994) When applied to landscapes this approachyields a form of knowledge characterized by (1) Time andtherefore history (2) Populations of landscape elements (egforest stands erosion source areas channel segments etc) (3)Processes and their interactions defined by frequency or prob-ability distributions and (4) Emergence of processes andpatterns at large scales that arise due to the collective behav-ior of large numbers of processes acting over smaller scales(Benda et al 1998) Several illustrative examples that exam-ine the effects of increasing scale in the watershed sciences aregiven below

Example One Forest Vegetation

Long-term (decadel to century) sequences of climate (rain-storms floods and fires) interacting with topography dictatepatterns of vegetation Distributions of climate topographyand vegetation are all scale dependent (eg Figures 1 and 2)

8 WMC Networker Summer 2001

In this example the scale dependency of vegetation is illus-trated using a coarse-grained fire model (Benda 1994) in thehumid temperate landscape of the Oregon Coast Range (OCR)Forest-replacing wildfires of a range of sizes (~1 ndash 1000 km2)over the last several millennia occurred at a mean interval ofabout 300 years although there was inter-millennial variabil-ity (Teensma 1987 Long 1995) In the OCR probabilitydistributions of forest ages are predicted to be positively

skewed and exponentially shaped (Benda 1994) (Figure 6)similar to fire model predictions in other landscapes (Johnsonand Van Wagener 1984) The right skewness arises because ofa decreasing probability of developing very old trees in conjunc-tion with an approximately equal susceptibility of fires acrossall age classes This forces the highest proportion of trees tooccur in the youngest age class and a gradual and systematicdecline in areas containing older trees

400 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

40 sq km

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

4000 sqkm

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

20000 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

0

005

01

015

02

025

03

0

200 400 600 800

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

035

50 250

450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650

850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

10 130

250

370

490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

012

014

016

10 130 250

370 490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

0

001

002

003

004

005

006

007

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

8 5

Figure 6 An example of scaling relationships using a forest fires and forest growth In the Oregon Coast Range fires occur onaverage once every 300 years and have sizes ranging from 1 to 1000 km2 (A) A simulated time series of mean forest age over a2500-year period showing increasing variability with decreasing spatial scale (B) Forest ages sampled in any one year will bebetter represented at larger spatial scales In addition the spatial distribution of ages at 20000 km2 should be statisticallysimilar to the long-term probability distribution of ages at any single point in the forest (C) The decreased variability of meanforest age with increasing spatial scale is represented in the form of a compressed and more symmetrical distribution ademonstration of the central limit theorem

WMC Networker Summer 2001 9

The forest fire example illustrates several of the scaling rela-tionships outlined earlier in this article Over sufficientlylarge areas the spatial distribution of forest ages measured inany year becomes similar to the expected right-skewed prob-ability distribution of ages at a single point in the forest(Figure 6) illustrating the ergodic hypothesis (Figure 5) How-ever the spatial size of the sample dictates the representationof that probability distribution illustrating a spatial scalingeffect (illustrated in Figure 2) When viewed as a populationof forest ages having mean value in each year the variabilityof the distribution of means decreases with increasing spatialscale with the distribution increasing in symmetry (Figure 6)

demonstrating the central limit theorem (Figure 3) Each ofthese patterns has geomorphological and ecological ramifica-tions In addition the increasing stability of the vegetationage distribution with increasing spatial scale has potentialimplications for characterizing environmental impacts (dis-cussed later)

Example Two Erosion by Landsliding

Erosion is dictated in large part by climate (ie rainstorms)slope steepness and vegetation Erosion often depends on athreshold being exceeded such as in rainfall-triggeredlandsliding (Caine 1990) or in sheetwash and gullying that is

km

5N

200 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

Tilton 200 km 2

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

esLandslides occur on average in 50 of every 100 years2 years out of every 100 have more than 10 landslides

km

5N

25 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10Number of Landslides

o

f Y

ears

Upper West Fork 25 km 2

Landslides occur on average in 15 of every 100 yearsOnly 3 of every 1000 year have more than 10 landslides

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

es

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

km

5N

3 km 2Upper West Fork 3 km 2

Landslides occur on average in only 4 of every 100 yearsand 3 of those 4 involve only 1 landslide

0

10

20

30

40

50

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f Lan

dsl

ides

a)

b)

c)

Figure 7The frequency of landsliding within a basin and the number of landslides occurring over any time varies systematicallywith basin size The 200 km2 area (a) encompasses thousands of potential landslide sites within this area landslides are relativelyfrequent and can occur in relatively large numbers The 25 km2 sub basin (b) contains fewer potential landslide sites landslidingwithin this area is less frequent with fewer landslides overall At 3 km2 (c) landsliding is very infrequent and any episode oflandsliding involves only a few sites

10 WMC Networker Summer 2001

dependent on soil hydrophobicity (Heede 1988) Punctuatederosion is ubiquitous in North America including in the south-ern coastal chaparral (Rice 1973) Cascade humid mountains(Swanson et al 1982) Pacific coastal rainforests (Hogan et al1995 Benda and Dunne 1997a) Appalachian Mountains (Hackand Goodlett 1960) and in the intermountain and highlandarid regions (Wohl and Pearthree 1991 Meyer et al 1995Robichaud and Brown 1999)

At the scale of populations of hillslopes over decades to centu-ries erosion occurs according to long-term patterns of climateinteracting with vegetation and topography To illustrate thiscentury-long patterns of erosion were investigated in the OCRusing a coarse-grained computational model (Benda and Dunne1997 Dunne 1998) In the simulation the temporal andspatial patterns of landsliding and debris flows are predictedto be an outcome of 1) probability distributions of stormintensity and duration 2) probability distributions of fire

recurrence and size 3) frequency distributions of topographicand geotechnical properties 4) deterministic thickening ofcolluvium in landslide sites 5) deterministic trajectories oftree rooting strength and 6) probability functions for sedi-ment transfer from hillslopes to channels The model pre-dicted that periodic fires and rainstorms triggered spates oflandsliding and debris flows every few decades to few centu-ries with little erosion at other times and places Low erosionrates punctuated by high magnitude releases of sediment atthe scale of a single hillslope or small basin is expressed by astrongly right-skewed or an exponential probability distribu-tion (Figure 7) Frequency of failures is low in small water-sheds because of the low frequency of fires and relatively smallnumber of landslide sites Landslide frequency and magni-tude increases with increasing drainage area because of in-creasing number of landslide sites and an increasing probabil-ity of storms and fires (Figure 7)

Example Three Channel Sedimentation

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

012345678910

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

00

05

10

15

0 500 1000 1500 2000 2500 3000

Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

FrequencyCumulative

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

180 km 2

45 km 2

25 km 2

00

05

10

15

20

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2) 3 km 2

Figure 8 An example of sediment fluctuations vary with spatial scale using computer simulations from southwest WashingtonThe probability distributions of fluxes evolve downstream to more symmetrical forms demonstrating the central limit theorem

WMC Networker Summer 2001 11

Over long periods channel networks obtain sediment at ratesdictated by probability distributions of erosion from indi-vidual hillslopes and small tributary basins (eg Figure 7)Sediment flux at any point in a channel network representstherefore an integrated sample of all upstream sources ofsupply Releases of sediment in a basin may be synchronousand correlated in time such as bank erosion during floods thatis likely to occur in smaller watersheds where storm size isequal to or greater than basin size Asynchronous erosion islikely the rule in basins where threshold-dependent erosionoccurs such as fire-induced surface erosion or storm triggeredmass wasting

A sediment routing model was applied to a mountain land-scape in southwest Washington to illustrate the effects ofsediment routing on the downstream distribution of sedimentflux The theoretical analysis required in addition to prob-ability distributions of erosion (eg Figure 7) probabilitydistributions of transport capacities transport velocities andparticle attrition (Benda and Dunne 1997b) Sediment fluxfrom individual hillslopes and small low-order basins is pre-

dicted to be punctaform and thereforestrongly right-skewed similar to theOCR example (Figure 7) The channelobtains sediment over long periodsfrom such numerous right-skewed dis-tributions in small headwater basinsThe accumulation of sediment fromthe multitude of sediment sources overtime causes the flux distribution toevolve into more symmetrical forms ademonstration of the central limittheorem (Figure 8 also see the moregeneral form in Figure 4) The modelalso predicts that the magnitude ofsediment perturbations decreasesdownstream in conjunction with a cor-responding increase in their numberIncrease in perturbation frequency isdue to an increasing number and there-fore probability of sediment releasesdownstream A decrease in magni-tude of the perturbation (ie differ-ence between minimum and maximumvalues) is due to several factors in-cluding 1) particle attrition (break-down) that increasingly damps sedi-ment signals with distance traveled2) a larger and persistent supply andtherefore store of gravel 3) diffusion ofdiscreet sediment pulses due to selec-tive transport and temporary storagein bars and 4) increasing channelwidth (Benda and Dunne 1997b)

Episodic introduction of sediment cancreates pulses or waves of sedimentthat migrate thought the network caus-ing changes in channel morphologyincluding variations in channel bedelevation sinuosity substrate sizespools etc (Gilbert 1917 Madej andOzake 1996 Benda 1990 Miller andBenda in press) In addition fluctua-tions in sediment supply may also

cause changes in sediment storage at a particular locationsuch as near fans or other low-gradient areas and these havebeen referred to as stationary waves (Benda and Dunne 1997b)Migrating or stationary waves create coarse-textured cut andfill terraces (Nakamura 1986 Roberts and Church 1986Miller and Benda in press) Hence the predicted fluctuationsin sediment supply by the model (Figure 8) have morphologicalconsequences to channels and valley floors and therefore on tothe biological communities that inhabit those environments

Example Four Large Woody Debris

The supply and storage of large woody debris (LWD) in streamsis governed by several landscape processes that occur punctu-ated in time over decades to centuries including by fireswindstorms landslides stand mortality and floods A woodbudgeting framework (Benda and Sias 1998 in press) wasused to evaluate how those landscape processes constrainlong-term patterns of wood abundance in streams

In the example below the effect of two end member fire

Rel

ativ

e F

requ

ency

(co

lum

ns)

Cum

ulative frequency (lines)

00

01

02

03

04

15 - 20 30 - 35 45 - 50 60 - 6500

04

06

08

150 500

150 500

Fire recurrence interval (years)

B

LWD storage (vu per 100 m reach)

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Year (first fire at year 500)

LWD

sto

rage

(vu

1

00 m

)A FIRES

(500-yr cycle)

Stand mortality

FIRES(150-yr cycle)

150500

Fire recurrence interval (years)

05 10

02

0 - 5 75 - 80

Figure 9 (A) Patterns of wood storage for fire cycles of 150 and 500 years Gradualincreases in storage represent chronic stand mortality The magnitude of the abruptpulses of wood storage is governed by the amount of standing biomass (controlled by forestage) and the time interval of the toppling of fire-killed trees (40 years in this example) (B)More frequent fires result in a compressed range of variability and a shift in thedistribution towards lower wood volumes (solid bars represent the 500-year cycle) Lessfrequent fires shift the distribution of LWD volumes towards the right into higher volumesThese patterns indicate the potential for significant differences in LWD storage alongclimatic gradients in the Pacific Northwest region (east to west and north to south)

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

4 WMC Networker Summer 2001

IntroductionMany environmental issues in the watershed sciences involv-ing forestry geology hydrology fisheries etc depend on inter-actions of numerous landscape processes acting over a largerange of spatial and temporal scales Consequently scientistsand managers have been seeking to understand the role ofscale in their respective disciplines for more than two decades(de Boer 1986 Allen and Starr 1982 Folt et al 1998) Forexample recognizing hierarchical time and space domains insurface processes brought issues of scale to the forefront in thescience of community and landscape ecology (Wiens 1984Frissel et al 1986) Motivated in part by ecologists watershedscientists also have recognized the need to transfer under-standing across spatial and temporal scales in the study oflandscapes (Swanson et al 1988 Benda et al 1998)

Despite this attention our understanding of how increasingthe dimensional scales of perception would play out in water-shed science management and regulation is limited Forexample it has proven difficult to extend channel reach-scalemeasurement and understanding obtained over a period of afew years to larger spatial and temporal scales (ie segmentto network over decades to centuries) because of un-reconciledstochastic processes and the lack of theory that would allowbridging between data gaps (Benda et al 1998) As a conse-quence most resource managers focus their planning activi-ties at small scales (harvest unit single rotation) with deci-sion making therefore hinged primarily on market forces andchanging environmental regulations Furthermore althoughregulators recognize the need for understanding resource man-agement and impact assessment at large scales they defaultto small scale reductionist approaches primarily because ofthe lack of leadership in this area by the scientific communityNevertheless this problem is on the radar screen as evidencedby a growing consensus calling for incorporating dynamics ordisturbance (ie increasing scale) into natural resource sci-ence and management (Botkin 1990 Naiman et al 1992Reeves et al 1995 Reid 1998 Dunne 1998 Lackey 1998Benda et al 1998 Maurer 1999) In addition adapting scienceto questions involving larger space and time scales is becom-ing a central tenet in new and creative forms of thinkingembodied in systems analysis (Weinberg 1975 Allen andStarr 1982) macro ecology (Mauer 1999) and complexityscience (Kellert 1993 Waldrop 1992 McIntrye 1998)

The purpose of this article is to outline highlight severaluniversal principles that address the effects of increasingscale in watersheds or landscapes Our aim is to equip readerswith a handful of tools that should allow them to think criti-cally regarding the role of scale in watershed science manage-ment and regulation in their home landscapes in the hope ofmoving beyond the hand waving stage In addition we empha-size how large-scale attributes of landscapes (ie mixtures ofclimate topography vegetation network geometry and basinscale) impose major constraints on the space and time struc-ture of environmental variability (ie the disturbance re-gime) embodied in forests erosion large organic debrisstreamflow channel conditions and aquatic habitats Be-cause the study of scale in the watershed sciences is a rela-tively new and rapidly evolving field we admit that under-

standing how increasing scale might play out in natural re-source management and regulation (forestry fisheries etc) isat an incubation stage We leave that topic up to the unbridledimagination of the reader

Universal Principles of ScaleA Landscape Parameter of Frequency and ProbabilityDistributions

Simply the system behavior of a landscape can be describedas the temporally changing values of an attribute at a singlelocation (forest ages sedimentation large woody debris etc)or the variation in these attributes over many locations in awatershed in a single year (Benda et al 1998) Hence thesystem behavior of landscapes cannot be understood solelywith single value parameters Landscape parameters mustbe multiple valued to reflect temporal variability and spatialheterogeneity key aspects to understanding geomorphic and

0

1

2

3

0 200 400 600 800 1000Years

Dep

th (

m) Temporal Sample

0

10

20

30

40

0 1 2 3Aggradation Depth (m)

Y

ears 100 yr

1000 yr

Spatial Sample

0

1

2

0 2 4 6 8 10

Distance Along Channel (km)

Dep

th (

m)

0

10

20

30

40

0 1 2Depth (m)

L

engt

h

1 km

10 km

Beyond Arm Waving Thinking Critically at Large ScalesContinued from Page 1

Figure 1 In temporally variable environments the longer thetime of observation the greater the probability of seeing alarge event here represented with a simulated time series ofsediment depth at a channel cross section Aggradationalevents create terraces with the highest terrace recording therarest events The frequency of aggradation influences thevegetation communities inhabiting each terrace

Figure 2 In spatially variable environments the larger thearea observed the greater is the probable range of observationsmade illustrated here with simulated sediment depths along a10-km channel reach

WMC Networker Summer 2001 5

ecologic behavior A distribution of values inexorably ariseswhen attributes are viewed either over long periods or largeregions of space it cannot be inferred from a point observationIn the parlance of the evolving field of complexity sciencefrequency or probability distributions comprise an emergentproperty that reveals the consequences of multiple interac-tions over multiple scales within landscapes

For example soil depth is a parameter and is used in manyapplications such as landslide prediction groundwater mod-eling etc The distribution of soil depth is however a param-eter itself because it contains information about the range andfrequency of values Hence when seeking to understand therole of scale distributions of parameters (such as soil depthforest ages rainfall etc) will be integrated into quantitativerelationships that predict watershed (system) behavior in theform of other distributions (Benda et al 1998) For examplethe probability distribution of climate coupled with frequencydistributions of topographic and vegetative attributes willyield a probability distribution of erosion and sediment flux(Benda and Dunne 1997) Distributions because they mea-sure temporal variability and spatial heterogeneity are sen-sitive to changing scales and they also indicate how measure-ment scale influences our perception of environments

Relationships Among Distributions Time and Space

Most physical and biological landscape processes are stronglydependent on scale For example the occurrence of distur-bances such as earthquakes landslides fires storms floodsetc is dependent on time scale These dynamic surface pro-cesses contributes to habitat diversity a multi-valued at-tribute that depends on spatial scale Hence intrinsic rela-tionships between temporal and spatial distributions andscale comprise a fundamental element for understanding howphysical and biological processes are interrelated Five uni-versal scaling relationships are briefly outlined below someare trivial and this section is meant as a review for manyreaders

Scaling in time (disturbance regime at a point) Manystochastic agents including earthquakes storms fires floodsand erosion act to create and modify channel and riparianlandforms The time required to observe the full range ofevents responsible for the full suite of riverine morphologies(eg disturbance regime or range of variability) depends on theoperative processes In general infrequent but large magni-tude events are responsible for large morphologic changes with

long legacies processes located in tails of probability distribu-tions (Dunne 1998) The structure of that temporal variabilityis increasingly apparent over larger time scales (Figure 1)

Scaling in space (many points diversity at a single time)An observer (human or otherwise) following a stream channelencounters a diversity of attributes (eg particle sizes pooldepths etc) that originate from temporal variability (Figure1) and deterministic topographic heterogeneity (ie system-atic decreases in gradient downstream for example) Therange of that diversity increases with the distance traversed(Figure 2) Thus the range of habitat types available to anorganism increases with increasing scale of movement Aquaticorganisms may have evolved different life history strategies in

10-kmChannel Network

1 km

02

04

06

08

0 25 50 75 100Year

Mea

nD

epth

(m)

0

10

20

30

0 05 1Depth (m)

L

engt

h

10 km

02

04

06

08

0 25 50 75 100Year

Mea

nD

epth

(m)

yr 40yr 45

0

10

20

0 05 1Depth (m)

L

engt

h yr 40

yr 45

1-km

reach

(A)

(B)

(A)

(B)

Sediment Depth

Y

ears

Sediment Depth

Y

ears

Sediment Depth

Y

earsChannel

Network

Sediment Source Area

Figure 4 Integration of multiple inputs via routing through achannel network causes the distribution of sediment flux andassociated storage to become more symmetrical with lowervariance downstream

Figure 3 In environments that vary in space and time thegreater the number of elements included in a population theless variable the distribution over time Here mean sedimentdepth over a 1-km length and a 10-km length of channel iscompared over a 100-yr time span Although the range ofvalues sampled over the 10-km length is greater the differenceis the distribution of values and in the mean value from yearto year is less

6 WMC Networker Summer 2001

response to the spatial and temporal scales of habitat hetero-geneity A similar pattern would be evident with landscapeattributes not linked to a channel network such as forestscomprised of different age stands or ages of landslide scars

Scaling in space and time (many points over time habi-tat stability) Stochastic agents of disturbance in combina-tion with time-invariant topographic features (ie channelgradients valley width etc) will change the set of attribute

values found at any single location over time (Figure 1) How-ever temporal variability can also be viewed at larger spatialscales say a population of hillslopes or channel reaches forexample The third universal relationship between distribu-tions and scale indicates that the extent a population ofsomething changes (ie for example channel reaches in termsof the shape of the distribution or of some statistical momentsuch as the mean mode or variance) is dependent on the size

150 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

500 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

200 300 400 500

150 years

0

2

4

6

8

10

12

Magnitude

Y

ea

rs

0 100 200 300 400 500

500 years

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

50 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

Time Series

0

100

200

300

400

500

600

700

800

0 100 200 300 400 500

Year

Mag

nitu

de

50 years

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

FIE

LD

ME

ASU

RE

ME

NT

S

MO

DE

L PR

ED

ICT

ION

Figure 5 At sufficiently large space and time scales the frequency distribution of a spatial ensemble of elements (Figure 2) maybecome statistically similar to a temporal distribution of that attribute at a single location (Figure 1) Similarity between largespatial samples and large temporal samples is referred to as an ergodic relationship (Hann 1977) and it may be used to infertemporal behavior from a spatial sample Although complete statistical equivalence is unlikely location-for-time substitution(Brundsen and Thornes 1978) allows for an understanding of temporal change from only a spatial sample

WMC Networker Summer 2001 7

of the population of elements Holding the mean size of adisturbance such as a fire or landslide constant the probabil-ity that the distribution of values will change over any giventime increases as the size of the population decreases (Figure3) This scaling relation is also observed in the temporalvariability of the mean value of the attribute under consider-ation This is simply a consequence that something (firelandslide flood) does not happen at the same intensity (oreven occur) every place at once The statistical stability of aspatial distribution of physical habitats (ie resilience toshifts over time) may influence such things as population sizespatial distribution and dispersal strategies

Scaling in time modulated by channel networks Scalingrelationships one through three are not only applicable tolandscapes or channel networks they are applicable to anyenvironment including galaxies or grains of sand on a beachThe fourth universal scaling relationship applies particularlyto branching networks such as a channel system Over longtime periods a channel network samples from a multitude ofindividual sources of mass flux (such as sediment and LWD)each with their own probability distribution (Figure 1) Theintegration of all sampled sources yields a long-term distribu-tion of mass flux through the network that evolves down-stream into more symmetrical forms a demonstration of thecentral limit theorem (Benda and Dunne 1997 Figure 4) Thisspatial evolution of the probability of channel and ripariandisturbances may impose spatial organization on aquatic andriparian habitats and on the species that inhabit them

Similarity Between Space and Time Variability (usingfield clues) At sufficiently large space and time scales thefrequency distribution of a spatial ensemble of elements (Fig-ure 2) may become statistically similar to a temporal distribu-tion of that attribute at a single location (Figure 1) thispotential commensurability is shown in Figure 5 For statis-tical similarity to hold all forcing functions must be tempo-rally stationary (over certain intervals) Similarity betweenlarge spatial samples and large temporal samples is referredto as an ergodic relationship (Hann 1977) and it may be usedto infer temporal behavior from a spatial sample Althoughcomplete statistical equivalence is unlikely location-for-timesubstitution (Brundsen and Thornes 1979) allows for an un-derstanding of temporal change from only a spatial sampleThis technique is particular useful for evaluating landscapebehavior including model predictions of same that encompassdecades to centuries during short human life spans Next weexamine some of the ramifications of increasing spatial andtemporal scale on watershed science management and regu-lation

Increasing Scale Implications for WatershedScienceCircumventing Limitations in Understanding CoarseGraining

Landscape interactions create patterns in the temporal be-havior and spatial distribution of watershed attributes thatmay be discernible only at large temporal and spatial scales(Figures 1 - 4) Because of the absence of long-term datacomputational models are needed to identify these patternsand the interactions that created them

Ideally large-scale computational models would be built uponphysics-based understanding of all relevant interdisciplinaryprocesses However there is a growing consensus that at leastin the near future there will be theoretical and technicalimpediments to developing large-scale environmental modelsbased on smaller-scale processes (Weinberg 1975 Dooge 1986)To circumvent present limitations watershed-scale modelscould be constructed whereby some small-scale processes areignored or are subsumed within larger-scale representationsof those processes a strategy commonly applied by physicistsand referred to as coarse graining (Gell-Mann 1994) In prac-tice in the watershed sciences this often requires combiningby means of mathematical synthesis and computer simula-tion empirical knowledge with theoretical reasoning avail-able at smaller scales to produce new understanding at largerscales (Roth et al 1989 Benda and Dunne 1997) The objec-tive of this approach would not be precise predictions aboutfuture states at individual sites but rather new testablehypotheses on large-scale interactions of climate topographyvegetation and riverine environments This approach has ahistory in the study of certain hydrological problems at largescales (Smith and Bretherton 1972 Rodriguez-Iturbe andValdez 1979)

Coarse-grained modeling is well suited to the use of distribu-tions and exploring the effects of scale on them (ie Figures 1ndash 5) System models by definition produce synthetic timeseries of watershed behavior (ie forest fires storms erosionchannel changes etc) In evaluating model predictions overhuman time frames (ie for hypothesis testing or environmen-tal problem solving) the typical absence of long-term dataforces us to consider a population of watershed elementssampled over a large spatial area Here again point observa-tions are insufficient to reveal system behavior and the distri-bution parameter in numerical models allow us to link behav-ior over large time and space scales to field observations madeover short times but large areas

Although the coarse-grained approach is taken by necessity inthe study of complex environmental systems there is consid-erable potential for obtaining new insights and understand-ing Focusing on small-scale processes often precludes seeingthe ldquobig picturerdquo that is the emergence of patterns and pro-cesses unseen at small scales Indeed the description of large-scale patterns of behavior even in the absence of mechanisticunderstanding for all of the observed processes is a hallmarkof the study of complex systems (Waldrop 1992 Kellert 1993Gell-Mann 1994) When applied to landscapes this approachyields a form of knowledge characterized by (1) Time andtherefore history (2) Populations of landscape elements (egforest stands erosion source areas channel segments etc) (3)Processes and their interactions defined by frequency or prob-ability distributions and (4) Emergence of processes andpatterns at large scales that arise due to the collective behav-ior of large numbers of processes acting over smaller scales(Benda et al 1998) Several illustrative examples that exam-ine the effects of increasing scale in the watershed sciences aregiven below

Example One Forest Vegetation

Long-term (decadel to century) sequences of climate (rain-storms floods and fires) interacting with topography dictatepatterns of vegetation Distributions of climate topographyand vegetation are all scale dependent (eg Figures 1 and 2)

8 WMC Networker Summer 2001

In this example the scale dependency of vegetation is illus-trated using a coarse-grained fire model (Benda 1994) in thehumid temperate landscape of the Oregon Coast Range (OCR)Forest-replacing wildfires of a range of sizes (~1 ndash 1000 km2)over the last several millennia occurred at a mean interval ofabout 300 years although there was inter-millennial variabil-ity (Teensma 1987 Long 1995) In the OCR probabilitydistributions of forest ages are predicted to be positively

skewed and exponentially shaped (Benda 1994) (Figure 6)similar to fire model predictions in other landscapes (Johnsonand Van Wagener 1984) The right skewness arises because ofa decreasing probability of developing very old trees in conjunc-tion with an approximately equal susceptibility of fires acrossall age classes This forces the highest proportion of trees tooccur in the youngest age class and a gradual and systematicdecline in areas containing older trees

400 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

40 sq km

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

4000 sqkm

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

20000 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

0

005

01

015

02

025

03

0

200 400 600 800

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

035

50 250

450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650

850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

10 130

250

370

490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

012

014

016

10 130 250

370 490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

0

001

002

003

004

005

006

007

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

8 5

Figure 6 An example of scaling relationships using a forest fires and forest growth In the Oregon Coast Range fires occur onaverage once every 300 years and have sizes ranging from 1 to 1000 km2 (A) A simulated time series of mean forest age over a2500-year period showing increasing variability with decreasing spatial scale (B) Forest ages sampled in any one year will bebetter represented at larger spatial scales In addition the spatial distribution of ages at 20000 km2 should be statisticallysimilar to the long-term probability distribution of ages at any single point in the forest (C) The decreased variability of meanforest age with increasing spatial scale is represented in the form of a compressed and more symmetrical distribution ademonstration of the central limit theorem

WMC Networker Summer 2001 9

The forest fire example illustrates several of the scaling rela-tionships outlined earlier in this article Over sufficientlylarge areas the spatial distribution of forest ages measured inany year becomes similar to the expected right-skewed prob-ability distribution of ages at a single point in the forest(Figure 6) illustrating the ergodic hypothesis (Figure 5) How-ever the spatial size of the sample dictates the representationof that probability distribution illustrating a spatial scalingeffect (illustrated in Figure 2) When viewed as a populationof forest ages having mean value in each year the variabilityof the distribution of means decreases with increasing spatialscale with the distribution increasing in symmetry (Figure 6)

demonstrating the central limit theorem (Figure 3) Each ofthese patterns has geomorphological and ecological ramifica-tions In addition the increasing stability of the vegetationage distribution with increasing spatial scale has potentialimplications for characterizing environmental impacts (dis-cussed later)

Example Two Erosion by Landsliding

Erosion is dictated in large part by climate (ie rainstorms)slope steepness and vegetation Erosion often depends on athreshold being exceeded such as in rainfall-triggeredlandsliding (Caine 1990) or in sheetwash and gullying that is

km

5N

200 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

Tilton 200 km 2

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

esLandslides occur on average in 50 of every 100 years2 years out of every 100 have more than 10 landslides

km

5N

25 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10Number of Landslides

o

f Y

ears

Upper West Fork 25 km 2

Landslides occur on average in 15 of every 100 yearsOnly 3 of every 1000 year have more than 10 landslides

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

es

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

km

5N

3 km 2Upper West Fork 3 km 2

Landslides occur on average in only 4 of every 100 yearsand 3 of those 4 involve only 1 landslide

0

10

20

30

40

50

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f Lan

dsl

ides

a)

b)

c)

Figure 7The frequency of landsliding within a basin and the number of landslides occurring over any time varies systematicallywith basin size The 200 km2 area (a) encompasses thousands of potential landslide sites within this area landslides are relativelyfrequent and can occur in relatively large numbers The 25 km2 sub basin (b) contains fewer potential landslide sites landslidingwithin this area is less frequent with fewer landslides overall At 3 km2 (c) landsliding is very infrequent and any episode oflandsliding involves only a few sites

10 WMC Networker Summer 2001

dependent on soil hydrophobicity (Heede 1988) Punctuatederosion is ubiquitous in North America including in the south-ern coastal chaparral (Rice 1973) Cascade humid mountains(Swanson et al 1982) Pacific coastal rainforests (Hogan et al1995 Benda and Dunne 1997a) Appalachian Mountains (Hackand Goodlett 1960) and in the intermountain and highlandarid regions (Wohl and Pearthree 1991 Meyer et al 1995Robichaud and Brown 1999)

At the scale of populations of hillslopes over decades to centu-ries erosion occurs according to long-term patterns of climateinteracting with vegetation and topography To illustrate thiscentury-long patterns of erosion were investigated in the OCRusing a coarse-grained computational model (Benda and Dunne1997 Dunne 1998) In the simulation the temporal andspatial patterns of landsliding and debris flows are predictedto be an outcome of 1) probability distributions of stormintensity and duration 2) probability distributions of fire

recurrence and size 3) frequency distributions of topographicand geotechnical properties 4) deterministic thickening ofcolluvium in landslide sites 5) deterministic trajectories oftree rooting strength and 6) probability functions for sedi-ment transfer from hillslopes to channels The model pre-dicted that periodic fires and rainstorms triggered spates oflandsliding and debris flows every few decades to few centu-ries with little erosion at other times and places Low erosionrates punctuated by high magnitude releases of sediment atthe scale of a single hillslope or small basin is expressed by astrongly right-skewed or an exponential probability distribu-tion (Figure 7) Frequency of failures is low in small water-sheds because of the low frequency of fires and relatively smallnumber of landslide sites Landslide frequency and magni-tude increases with increasing drainage area because of in-creasing number of landslide sites and an increasing probabil-ity of storms and fires (Figure 7)

Example Three Channel Sedimentation

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

012345678910

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

00

05

10

15

0 500 1000 1500 2000 2500 3000

Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

FrequencyCumulative

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

180 km 2

45 km 2

25 km 2

00

05

10

15

20

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2) 3 km 2

Figure 8 An example of sediment fluctuations vary with spatial scale using computer simulations from southwest WashingtonThe probability distributions of fluxes evolve downstream to more symmetrical forms demonstrating the central limit theorem

WMC Networker Summer 2001 11

Over long periods channel networks obtain sediment at ratesdictated by probability distributions of erosion from indi-vidual hillslopes and small tributary basins (eg Figure 7)Sediment flux at any point in a channel network representstherefore an integrated sample of all upstream sources ofsupply Releases of sediment in a basin may be synchronousand correlated in time such as bank erosion during floods thatis likely to occur in smaller watersheds where storm size isequal to or greater than basin size Asynchronous erosion islikely the rule in basins where threshold-dependent erosionoccurs such as fire-induced surface erosion or storm triggeredmass wasting

A sediment routing model was applied to a mountain land-scape in southwest Washington to illustrate the effects ofsediment routing on the downstream distribution of sedimentflux The theoretical analysis required in addition to prob-ability distributions of erosion (eg Figure 7) probabilitydistributions of transport capacities transport velocities andparticle attrition (Benda and Dunne 1997b) Sediment fluxfrom individual hillslopes and small low-order basins is pre-

dicted to be punctaform and thereforestrongly right-skewed similar to theOCR example (Figure 7) The channelobtains sediment over long periodsfrom such numerous right-skewed dis-tributions in small headwater basinsThe accumulation of sediment fromthe multitude of sediment sources overtime causes the flux distribution toevolve into more symmetrical forms ademonstration of the central limittheorem (Figure 8 also see the moregeneral form in Figure 4) The modelalso predicts that the magnitude ofsediment perturbations decreasesdownstream in conjunction with a cor-responding increase in their numberIncrease in perturbation frequency isdue to an increasing number and there-fore probability of sediment releasesdownstream A decrease in magni-tude of the perturbation (ie differ-ence between minimum and maximumvalues) is due to several factors in-cluding 1) particle attrition (break-down) that increasingly damps sedi-ment signals with distance traveled2) a larger and persistent supply andtherefore store of gravel 3) diffusion ofdiscreet sediment pulses due to selec-tive transport and temporary storagein bars and 4) increasing channelwidth (Benda and Dunne 1997b)

Episodic introduction of sediment cancreates pulses or waves of sedimentthat migrate thought the network caus-ing changes in channel morphologyincluding variations in channel bedelevation sinuosity substrate sizespools etc (Gilbert 1917 Madej andOzake 1996 Benda 1990 Miller andBenda in press) In addition fluctua-tions in sediment supply may also

cause changes in sediment storage at a particular locationsuch as near fans or other low-gradient areas and these havebeen referred to as stationary waves (Benda and Dunne 1997b)Migrating or stationary waves create coarse-textured cut andfill terraces (Nakamura 1986 Roberts and Church 1986Miller and Benda in press) Hence the predicted fluctuationsin sediment supply by the model (Figure 8) have morphologicalconsequences to channels and valley floors and therefore on tothe biological communities that inhabit those environments

Example Four Large Woody Debris

The supply and storage of large woody debris (LWD) in streamsis governed by several landscape processes that occur punctu-ated in time over decades to centuries including by fireswindstorms landslides stand mortality and floods A woodbudgeting framework (Benda and Sias 1998 in press) wasused to evaluate how those landscape processes constrainlong-term patterns of wood abundance in streams

In the example below the effect of two end member fire

Rel

ativ

e F

requ

ency

(co

lum

ns)

Cum

ulative frequency (lines)

00

01

02

03

04

15 - 20 30 - 35 45 - 50 60 - 6500

04

06

08

150 500

150 500

Fire recurrence interval (years)

B

LWD storage (vu per 100 m reach)

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Year (first fire at year 500)

LWD

sto

rage

(vu

1

00 m

)A FIRES

(500-yr cycle)

Stand mortality

FIRES(150-yr cycle)

150500

Fire recurrence interval (years)

05 10

02

0 - 5 75 - 80

Figure 9 (A) Patterns of wood storage for fire cycles of 150 and 500 years Gradualincreases in storage represent chronic stand mortality The magnitude of the abruptpulses of wood storage is governed by the amount of standing biomass (controlled by forestage) and the time interval of the toppling of fire-killed trees (40 years in this example) (B)More frequent fires result in a compressed range of variability and a shift in thedistribution towards lower wood volumes (solid bars represent the 500-year cycle) Lessfrequent fires shift the distribution of LWD volumes towards the right into higher volumesThese patterns indicate the potential for significant differences in LWD storage alongclimatic gradients in the Pacific Northwest region (east to west and north to south)

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 5

ecologic behavior A distribution of values inexorably ariseswhen attributes are viewed either over long periods or largeregions of space it cannot be inferred from a point observationIn the parlance of the evolving field of complexity sciencefrequency or probability distributions comprise an emergentproperty that reveals the consequences of multiple interac-tions over multiple scales within landscapes

For example soil depth is a parameter and is used in manyapplications such as landslide prediction groundwater mod-eling etc The distribution of soil depth is however a param-eter itself because it contains information about the range andfrequency of values Hence when seeking to understand therole of scale distributions of parameters (such as soil depthforest ages rainfall etc) will be integrated into quantitativerelationships that predict watershed (system) behavior in theform of other distributions (Benda et al 1998) For examplethe probability distribution of climate coupled with frequencydistributions of topographic and vegetative attributes willyield a probability distribution of erosion and sediment flux(Benda and Dunne 1997) Distributions because they mea-sure temporal variability and spatial heterogeneity are sen-sitive to changing scales and they also indicate how measure-ment scale influences our perception of environments

Relationships Among Distributions Time and Space

Most physical and biological landscape processes are stronglydependent on scale For example the occurrence of distur-bances such as earthquakes landslides fires storms floodsetc is dependent on time scale These dynamic surface pro-cesses contributes to habitat diversity a multi-valued at-tribute that depends on spatial scale Hence intrinsic rela-tionships between temporal and spatial distributions andscale comprise a fundamental element for understanding howphysical and biological processes are interrelated Five uni-versal scaling relationships are briefly outlined below someare trivial and this section is meant as a review for manyreaders

Scaling in time (disturbance regime at a point) Manystochastic agents including earthquakes storms fires floodsand erosion act to create and modify channel and riparianlandforms The time required to observe the full range ofevents responsible for the full suite of riverine morphologies(eg disturbance regime or range of variability) depends on theoperative processes In general infrequent but large magni-tude events are responsible for large morphologic changes with

long legacies processes located in tails of probability distribu-tions (Dunne 1998) The structure of that temporal variabilityis increasingly apparent over larger time scales (Figure 1)

Scaling in space (many points diversity at a single time)An observer (human or otherwise) following a stream channelencounters a diversity of attributes (eg particle sizes pooldepths etc) that originate from temporal variability (Figure1) and deterministic topographic heterogeneity (ie system-atic decreases in gradient downstream for example) Therange of that diversity increases with the distance traversed(Figure 2) Thus the range of habitat types available to anorganism increases with increasing scale of movement Aquaticorganisms may have evolved different life history strategies in

10-kmChannel Network

1 km

02

04

06

08

0 25 50 75 100Year

Mea

nD

epth

(m)

0

10

20

30

0 05 1Depth (m)

L

engt

h

10 km

02

04

06

08

0 25 50 75 100Year

Mea

nD

epth

(m)

yr 40yr 45

0

10

20

0 05 1Depth (m)

L

engt

h yr 40

yr 45

1-km

reach

(A)

(B)

(A)

(B)

Sediment Depth

Y

ears

Sediment Depth

Y

ears

Sediment Depth

Y

earsChannel

Network

Sediment Source Area

Figure 4 Integration of multiple inputs via routing through achannel network causes the distribution of sediment flux andassociated storage to become more symmetrical with lowervariance downstream

Figure 3 In environments that vary in space and time thegreater the number of elements included in a population theless variable the distribution over time Here mean sedimentdepth over a 1-km length and a 10-km length of channel iscompared over a 100-yr time span Although the range ofvalues sampled over the 10-km length is greater the differenceis the distribution of values and in the mean value from yearto year is less

6 WMC Networker Summer 2001

response to the spatial and temporal scales of habitat hetero-geneity A similar pattern would be evident with landscapeattributes not linked to a channel network such as forestscomprised of different age stands or ages of landslide scars

Scaling in space and time (many points over time habi-tat stability) Stochastic agents of disturbance in combina-tion with time-invariant topographic features (ie channelgradients valley width etc) will change the set of attribute

values found at any single location over time (Figure 1) How-ever temporal variability can also be viewed at larger spatialscales say a population of hillslopes or channel reaches forexample The third universal relationship between distribu-tions and scale indicates that the extent a population ofsomething changes (ie for example channel reaches in termsof the shape of the distribution or of some statistical momentsuch as the mean mode or variance) is dependent on the size

150 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

500 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

200 300 400 500

150 years

0

2

4

6

8

10

12

Magnitude

Y

ea

rs

0 100 200 300 400 500

500 years

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

50 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

Time Series

0

100

200

300

400

500

600

700

800

0 100 200 300 400 500

Year

Mag

nitu

de

50 years

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

FIE

LD

ME

ASU

RE

ME

NT

S

MO

DE

L PR

ED

ICT

ION

Figure 5 At sufficiently large space and time scales the frequency distribution of a spatial ensemble of elements (Figure 2) maybecome statistically similar to a temporal distribution of that attribute at a single location (Figure 1) Similarity between largespatial samples and large temporal samples is referred to as an ergodic relationship (Hann 1977) and it may be used to infertemporal behavior from a spatial sample Although complete statistical equivalence is unlikely location-for-time substitution(Brundsen and Thornes 1978) allows for an understanding of temporal change from only a spatial sample

WMC Networker Summer 2001 7

of the population of elements Holding the mean size of adisturbance such as a fire or landslide constant the probabil-ity that the distribution of values will change over any giventime increases as the size of the population decreases (Figure3) This scaling relation is also observed in the temporalvariability of the mean value of the attribute under consider-ation This is simply a consequence that something (firelandslide flood) does not happen at the same intensity (oreven occur) every place at once The statistical stability of aspatial distribution of physical habitats (ie resilience toshifts over time) may influence such things as population sizespatial distribution and dispersal strategies

Scaling in time modulated by channel networks Scalingrelationships one through three are not only applicable tolandscapes or channel networks they are applicable to anyenvironment including galaxies or grains of sand on a beachThe fourth universal scaling relationship applies particularlyto branching networks such as a channel system Over longtime periods a channel network samples from a multitude ofindividual sources of mass flux (such as sediment and LWD)each with their own probability distribution (Figure 1) Theintegration of all sampled sources yields a long-term distribu-tion of mass flux through the network that evolves down-stream into more symmetrical forms a demonstration of thecentral limit theorem (Benda and Dunne 1997 Figure 4) Thisspatial evolution of the probability of channel and ripariandisturbances may impose spatial organization on aquatic andriparian habitats and on the species that inhabit them

Similarity Between Space and Time Variability (usingfield clues) At sufficiently large space and time scales thefrequency distribution of a spatial ensemble of elements (Fig-ure 2) may become statistically similar to a temporal distribu-tion of that attribute at a single location (Figure 1) thispotential commensurability is shown in Figure 5 For statis-tical similarity to hold all forcing functions must be tempo-rally stationary (over certain intervals) Similarity betweenlarge spatial samples and large temporal samples is referredto as an ergodic relationship (Hann 1977) and it may be usedto infer temporal behavior from a spatial sample Althoughcomplete statistical equivalence is unlikely location-for-timesubstitution (Brundsen and Thornes 1979) allows for an un-derstanding of temporal change from only a spatial sampleThis technique is particular useful for evaluating landscapebehavior including model predictions of same that encompassdecades to centuries during short human life spans Next weexamine some of the ramifications of increasing spatial andtemporal scale on watershed science management and regu-lation

Increasing Scale Implications for WatershedScienceCircumventing Limitations in Understanding CoarseGraining

Landscape interactions create patterns in the temporal be-havior and spatial distribution of watershed attributes thatmay be discernible only at large temporal and spatial scales(Figures 1 - 4) Because of the absence of long-term datacomputational models are needed to identify these patternsand the interactions that created them

Ideally large-scale computational models would be built uponphysics-based understanding of all relevant interdisciplinaryprocesses However there is a growing consensus that at leastin the near future there will be theoretical and technicalimpediments to developing large-scale environmental modelsbased on smaller-scale processes (Weinberg 1975 Dooge 1986)To circumvent present limitations watershed-scale modelscould be constructed whereby some small-scale processes areignored or are subsumed within larger-scale representationsof those processes a strategy commonly applied by physicistsand referred to as coarse graining (Gell-Mann 1994) In prac-tice in the watershed sciences this often requires combiningby means of mathematical synthesis and computer simula-tion empirical knowledge with theoretical reasoning avail-able at smaller scales to produce new understanding at largerscales (Roth et al 1989 Benda and Dunne 1997) The objec-tive of this approach would not be precise predictions aboutfuture states at individual sites but rather new testablehypotheses on large-scale interactions of climate topographyvegetation and riverine environments This approach has ahistory in the study of certain hydrological problems at largescales (Smith and Bretherton 1972 Rodriguez-Iturbe andValdez 1979)

Coarse-grained modeling is well suited to the use of distribu-tions and exploring the effects of scale on them (ie Figures 1ndash 5) System models by definition produce synthetic timeseries of watershed behavior (ie forest fires storms erosionchannel changes etc) In evaluating model predictions overhuman time frames (ie for hypothesis testing or environmen-tal problem solving) the typical absence of long-term dataforces us to consider a population of watershed elementssampled over a large spatial area Here again point observa-tions are insufficient to reveal system behavior and the distri-bution parameter in numerical models allow us to link behav-ior over large time and space scales to field observations madeover short times but large areas

Although the coarse-grained approach is taken by necessity inthe study of complex environmental systems there is consid-erable potential for obtaining new insights and understand-ing Focusing on small-scale processes often precludes seeingthe ldquobig picturerdquo that is the emergence of patterns and pro-cesses unseen at small scales Indeed the description of large-scale patterns of behavior even in the absence of mechanisticunderstanding for all of the observed processes is a hallmarkof the study of complex systems (Waldrop 1992 Kellert 1993Gell-Mann 1994) When applied to landscapes this approachyields a form of knowledge characterized by (1) Time andtherefore history (2) Populations of landscape elements (egforest stands erosion source areas channel segments etc) (3)Processes and their interactions defined by frequency or prob-ability distributions and (4) Emergence of processes andpatterns at large scales that arise due to the collective behav-ior of large numbers of processes acting over smaller scales(Benda et al 1998) Several illustrative examples that exam-ine the effects of increasing scale in the watershed sciences aregiven below

Example One Forest Vegetation

Long-term (decadel to century) sequences of climate (rain-storms floods and fires) interacting with topography dictatepatterns of vegetation Distributions of climate topographyand vegetation are all scale dependent (eg Figures 1 and 2)

8 WMC Networker Summer 2001

In this example the scale dependency of vegetation is illus-trated using a coarse-grained fire model (Benda 1994) in thehumid temperate landscape of the Oregon Coast Range (OCR)Forest-replacing wildfires of a range of sizes (~1 ndash 1000 km2)over the last several millennia occurred at a mean interval ofabout 300 years although there was inter-millennial variabil-ity (Teensma 1987 Long 1995) In the OCR probabilitydistributions of forest ages are predicted to be positively

skewed and exponentially shaped (Benda 1994) (Figure 6)similar to fire model predictions in other landscapes (Johnsonand Van Wagener 1984) The right skewness arises because ofa decreasing probability of developing very old trees in conjunc-tion with an approximately equal susceptibility of fires acrossall age classes This forces the highest proportion of trees tooccur in the youngest age class and a gradual and systematicdecline in areas containing older trees

400 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

40 sq km

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

4000 sqkm

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

20000 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

0

005

01

015

02

025

03

0

200 400 600 800

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

035

50 250

450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650

850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

10 130

250

370

490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

012

014

016

10 130 250

370 490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

0

001

002

003

004

005

006

007

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

8 5

Figure 6 An example of scaling relationships using a forest fires and forest growth In the Oregon Coast Range fires occur onaverage once every 300 years and have sizes ranging from 1 to 1000 km2 (A) A simulated time series of mean forest age over a2500-year period showing increasing variability with decreasing spatial scale (B) Forest ages sampled in any one year will bebetter represented at larger spatial scales In addition the spatial distribution of ages at 20000 km2 should be statisticallysimilar to the long-term probability distribution of ages at any single point in the forest (C) The decreased variability of meanforest age with increasing spatial scale is represented in the form of a compressed and more symmetrical distribution ademonstration of the central limit theorem

WMC Networker Summer 2001 9

The forest fire example illustrates several of the scaling rela-tionships outlined earlier in this article Over sufficientlylarge areas the spatial distribution of forest ages measured inany year becomes similar to the expected right-skewed prob-ability distribution of ages at a single point in the forest(Figure 6) illustrating the ergodic hypothesis (Figure 5) How-ever the spatial size of the sample dictates the representationof that probability distribution illustrating a spatial scalingeffect (illustrated in Figure 2) When viewed as a populationof forest ages having mean value in each year the variabilityof the distribution of means decreases with increasing spatialscale with the distribution increasing in symmetry (Figure 6)

demonstrating the central limit theorem (Figure 3) Each ofthese patterns has geomorphological and ecological ramifica-tions In addition the increasing stability of the vegetationage distribution with increasing spatial scale has potentialimplications for characterizing environmental impacts (dis-cussed later)

Example Two Erosion by Landsliding

Erosion is dictated in large part by climate (ie rainstorms)slope steepness and vegetation Erosion often depends on athreshold being exceeded such as in rainfall-triggeredlandsliding (Caine 1990) or in sheetwash and gullying that is

km

5N

200 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

Tilton 200 km 2

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

esLandslides occur on average in 50 of every 100 years2 years out of every 100 have more than 10 landslides

km

5N

25 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10Number of Landslides

o

f Y

ears

Upper West Fork 25 km 2

Landslides occur on average in 15 of every 100 yearsOnly 3 of every 1000 year have more than 10 landslides

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

es

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

km

5N

3 km 2Upper West Fork 3 km 2

Landslides occur on average in only 4 of every 100 yearsand 3 of those 4 involve only 1 landslide

0

10

20

30

40

50

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f Lan

dsl

ides

a)

b)

c)

Figure 7The frequency of landsliding within a basin and the number of landslides occurring over any time varies systematicallywith basin size The 200 km2 area (a) encompasses thousands of potential landslide sites within this area landslides are relativelyfrequent and can occur in relatively large numbers The 25 km2 sub basin (b) contains fewer potential landslide sites landslidingwithin this area is less frequent with fewer landslides overall At 3 km2 (c) landsliding is very infrequent and any episode oflandsliding involves only a few sites

10 WMC Networker Summer 2001

dependent on soil hydrophobicity (Heede 1988) Punctuatederosion is ubiquitous in North America including in the south-ern coastal chaparral (Rice 1973) Cascade humid mountains(Swanson et al 1982) Pacific coastal rainforests (Hogan et al1995 Benda and Dunne 1997a) Appalachian Mountains (Hackand Goodlett 1960) and in the intermountain and highlandarid regions (Wohl and Pearthree 1991 Meyer et al 1995Robichaud and Brown 1999)

At the scale of populations of hillslopes over decades to centu-ries erosion occurs according to long-term patterns of climateinteracting with vegetation and topography To illustrate thiscentury-long patterns of erosion were investigated in the OCRusing a coarse-grained computational model (Benda and Dunne1997 Dunne 1998) In the simulation the temporal andspatial patterns of landsliding and debris flows are predictedto be an outcome of 1) probability distributions of stormintensity and duration 2) probability distributions of fire

recurrence and size 3) frequency distributions of topographicand geotechnical properties 4) deterministic thickening ofcolluvium in landslide sites 5) deterministic trajectories oftree rooting strength and 6) probability functions for sedi-ment transfer from hillslopes to channels The model pre-dicted that periodic fires and rainstorms triggered spates oflandsliding and debris flows every few decades to few centu-ries with little erosion at other times and places Low erosionrates punctuated by high magnitude releases of sediment atthe scale of a single hillslope or small basin is expressed by astrongly right-skewed or an exponential probability distribu-tion (Figure 7) Frequency of failures is low in small water-sheds because of the low frequency of fires and relatively smallnumber of landslide sites Landslide frequency and magni-tude increases with increasing drainage area because of in-creasing number of landslide sites and an increasing probabil-ity of storms and fires (Figure 7)

Example Three Channel Sedimentation

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

012345678910

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

00

05

10

15

0 500 1000 1500 2000 2500 3000

Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

FrequencyCumulative

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

180 km 2

45 km 2

25 km 2

00

05

10

15

20

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2) 3 km 2

Figure 8 An example of sediment fluctuations vary with spatial scale using computer simulations from southwest WashingtonThe probability distributions of fluxes evolve downstream to more symmetrical forms demonstrating the central limit theorem

WMC Networker Summer 2001 11

Over long periods channel networks obtain sediment at ratesdictated by probability distributions of erosion from indi-vidual hillslopes and small tributary basins (eg Figure 7)Sediment flux at any point in a channel network representstherefore an integrated sample of all upstream sources ofsupply Releases of sediment in a basin may be synchronousand correlated in time such as bank erosion during floods thatis likely to occur in smaller watersheds where storm size isequal to or greater than basin size Asynchronous erosion islikely the rule in basins where threshold-dependent erosionoccurs such as fire-induced surface erosion or storm triggeredmass wasting

A sediment routing model was applied to a mountain land-scape in southwest Washington to illustrate the effects ofsediment routing on the downstream distribution of sedimentflux The theoretical analysis required in addition to prob-ability distributions of erosion (eg Figure 7) probabilitydistributions of transport capacities transport velocities andparticle attrition (Benda and Dunne 1997b) Sediment fluxfrom individual hillslopes and small low-order basins is pre-

dicted to be punctaform and thereforestrongly right-skewed similar to theOCR example (Figure 7) The channelobtains sediment over long periodsfrom such numerous right-skewed dis-tributions in small headwater basinsThe accumulation of sediment fromthe multitude of sediment sources overtime causes the flux distribution toevolve into more symmetrical forms ademonstration of the central limittheorem (Figure 8 also see the moregeneral form in Figure 4) The modelalso predicts that the magnitude ofsediment perturbations decreasesdownstream in conjunction with a cor-responding increase in their numberIncrease in perturbation frequency isdue to an increasing number and there-fore probability of sediment releasesdownstream A decrease in magni-tude of the perturbation (ie differ-ence between minimum and maximumvalues) is due to several factors in-cluding 1) particle attrition (break-down) that increasingly damps sedi-ment signals with distance traveled2) a larger and persistent supply andtherefore store of gravel 3) diffusion ofdiscreet sediment pulses due to selec-tive transport and temporary storagein bars and 4) increasing channelwidth (Benda and Dunne 1997b)

Episodic introduction of sediment cancreates pulses or waves of sedimentthat migrate thought the network caus-ing changes in channel morphologyincluding variations in channel bedelevation sinuosity substrate sizespools etc (Gilbert 1917 Madej andOzake 1996 Benda 1990 Miller andBenda in press) In addition fluctua-tions in sediment supply may also

cause changes in sediment storage at a particular locationsuch as near fans or other low-gradient areas and these havebeen referred to as stationary waves (Benda and Dunne 1997b)Migrating or stationary waves create coarse-textured cut andfill terraces (Nakamura 1986 Roberts and Church 1986Miller and Benda in press) Hence the predicted fluctuationsin sediment supply by the model (Figure 8) have morphologicalconsequences to channels and valley floors and therefore on tothe biological communities that inhabit those environments

Example Four Large Woody Debris

The supply and storage of large woody debris (LWD) in streamsis governed by several landscape processes that occur punctu-ated in time over decades to centuries including by fireswindstorms landslides stand mortality and floods A woodbudgeting framework (Benda and Sias 1998 in press) wasused to evaluate how those landscape processes constrainlong-term patterns of wood abundance in streams

In the example below the effect of two end member fire

Rel

ativ

e F

requ

ency

(co

lum

ns)

Cum

ulative frequency (lines)

00

01

02

03

04

15 - 20 30 - 35 45 - 50 60 - 6500

04

06

08

150 500

150 500

Fire recurrence interval (years)

B

LWD storage (vu per 100 m reach)

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Year (first fire at year 500)

LWD

sto

rage

(vu

1

00 m

)A FIRES

(500-yr cycle)

Stand mortality

FIRES(150-yr cycle)

150500

Fire recurrence interval (years)

05 10

02

0 - 5 75 - 80

Figure 9 (A) Patterns of wood storage for fire cycles of 150 and 500 years Gradualincreases in storage represent chronic stand mortality The magnitude of the abruptpulses of wood storage is governed by the amount of standing biomass (controlled by forestage) and the time interval of the toppling of fire-killed trees (40 years in this example) (B)More frequent fires result in a compressed range of variability and a shift in thedistribution towards lower wood volumes (solid bars represent the 500-year cycle) Lessfrequent fires shift the distribution of LWD volumes towards the right into higher volumesThese patterns indicate the potential for significant differences in LWD storage alongclimatic gradients in the Pacific Northwest region (east to west and north to south)

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

6 WMC Networker Summer 2001

response to the spatial and temporal scales of habitat hetero-geneity A similar pattern would be evident with landscapeattributes not linked to a channel network such as forestscomprised of different age stands or ages of landslide scars

Scaling in space and time (many points over time habi-tat stability) Stochastic agents of disturbance in combina-tion with time-invariant topographic features (ie channelgradients valley width etc) will change the set of attribute

values found at any single location over time (Figure 1) How-ever temporal variability can also be viewed at larger spatialscales say a population of hillslopes or channel reaches forexample The third universal relationship between distribu-tions and scale indicates that the extent a population ofsomething changes (ie for example channel reaches in termsof the shape of the distribution or of some statistical momentsuch as the mean mode or variance) is dependent on the size

150 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

500 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

200 300 400 500

150 years

0

2

4

6

8

10

12

Magnitude

Y

ea

rs

0 100 200 300 400 500

500 years

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

50 sites

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

Time Series

0

100

200

300

400

500

600

700

800

0 100 200 300 400 500

Year

Mag

nitu

de

50 years

0

2

4

6

8

10

12

Magnitude

Y

ears

0 100 200 300 400 500

FIE

LD

ME

ASU

RE

ME

NT

S

MO

DE

L PR

ED

ICT

ION

Figure 5 At sufficiently large space and time scales the frequency distribution of a spatial ensemble of elements (Figure 2) maybecome statistically similar to a temporal distribution of that attribute at a single location (Figure 1) Similarity between largespatial samples and large temporal samples is referred to as an ergodic relationship (Hann 1977) and it may be used to infertemporal behavior from a spatial sample Although complete statistical equivalence is unlikely location-for-time substitution(Brundsen and Thornes 1978) allows for an understanding of temporal change from only a spatial sample

WMC Networker Summer 2001 7

of the population of elements Holding the mean size of adisturbance such as a fire or landslide constant the probabil-ity that the distribution of values will change over any giventime increases as the size of the population decreases (Figure3) This scaling relation is also observed in the temporalvariability of the mean value of the attribute under consider-ation This is simply a consequence that something (firelandslide flood) does not happen at the same intensity (oreven occur) every place at once The statistical stability of aspatial distribution of physical habitats (ie resilience toshifts over time) may influence such things as population sizespatial distribution and dispersal strategies

Scaling in time modulated by channel networks Scalingrelationships one through three are not only applicable tolandscapes or channel networks they are applicable to anyenvironment including galaxies or grains of sand on a beachThe fourth universal scaling relationship applies particularlyto branching networks such as a channel system Over longtime periods a channel network samples from a multitude ofindividual sources of mass flux (such as sediment and LWD)each with their own probability distribution (Figure 1) Theintegration of all sampled sources yields a long-term distribu-tion of mass flux through the network that evolves down-stream into more symmetrical forms a demonstration of thecentral limit theorem (Benda and Dunne 1997 Figure 4) Thisspatial evolution of the probability of channel and ripariandisturbances may impose spatial organization on aquatic andriparian habitats and on the species that inhabit them

Similarity Between Space and Time Variability (usingfield clues) At sufficiently large space and time scales thefrequency distribution of a spatial ensemble of elements (Fig-ure 2) may become statistically similar to a temporal distribu-tion of that attribute at a single location (Figure 1) thispotential commensurability is shown in Figure 5 For statis-tical similarity to hold all forcing functions must be tempo-rally stationary (over certain intervals) Similarity betweenlarge spatial samples and large temporal samples is referredto as an ergodic relationship (Hann 1977) and it may be usedto infer temporal behavior from a spatial sample Althoughcomplete statistical equivalence is unlikely location-for-timesubstitution (Brundsen and Thornes 1979) allows for an un-derstanding of temporal change from only a spatial sampleThis technique is particular useful for evaluating landscapebehavior including model predictions of same that encompassdecades to centuries during short human life spans Next weexamine some of the ramifications of increasing spatial andtemporal scale on watershed science management and regu-lation

Increasing Scale Implications for WatershedScienceCircumventing Limitations in Understanding CoarseGraining

Landscape interactions create patterns in the temporal be-havior and spatial distribution of watershed attributes thatmay be discernible only at large temporal and spatial scales(Figures 1 - 4) Because of the absence of long-term datacomputational models are needed to identify these patternsand the interactions that created them

Ideally large-scale computational models would be built uponphysics-based understanding of all relevant interdisciplinaryprocesses However there is a growing consensus that at leastin the near future there will be theoretical and technicalimpediments to developing large-scale environmental modelsbased on smaller-scale processes (Weinberg 1975 Dooge 1986)To circumvent present limitations watershed-scale modelscould be constructed whereby some small-scale processes areignored or are subsumed within larger-scale representationsof those processes a strategy commonly applied by physicistsand referred to as coarse graining (Gell-Mann 1994) In prac-tice in the watershed sciences this often requires combiningby means of mathematical synthesis and computer simula-tion empirical knowledge with theoretical reasoning avail-able at smaller scales to produce new understanding at largerscales (Roth et al 1989 Benda and Dunne 1997) The objec-tive of this approach would not be precise predictions aboutfuture states at individual sites but rather new testablehypotheses on large-scale interactions of climate topographyvegetation and riverine environments This approach has ahistory in the study of certain hydrological problems at largescales (Smith and Bretherton 1972 Rodriguez-Iturbe andValdez 1979)

Coarse-grained modeling is well suited to the use of distribu-tions and exploring the effects of scale on them (ie Figures 1ndash 5) System models by definition produce synthetic timeseries of watershed behavior (ie forest fires storms erosionchannel changes etc) In evaluating model predictions overhuman time frames (ie for hypothesis testing or environmen-tal problem solving) the typical absence of long-term dataforces us to consider a population of watershed elementssampled over a large spatial area Here again point observa-tions are insufficient to reveal system behavior and the distri-bution parameter in numerical models allow us to link behav-ior over large time and space scales to field observations madeover short times but large areas

Although the coarse-grained approach is taken by necessity inthe study of complex environmental systems there is consid-erable potential for obtaining new insights and understand-ing Focusing on small-scale processes often precludes seeingthe ldquobig picturerdquo that is the emergence of patterns and pro-cesses unseen at small scales Indeed the description of large-scale patterns of behavior even in the absence of mechanisticunderstanding for all of the observed processes is a hallmarkof the study of complex systems (Waldrop 1992 Kellert 1993Gell-Mann 1994) When applied to landscapes this approachyields a form of knowledge characterized by (1) Time andtherefore history (2) Populations of landscape elements (egforest stands erosion source areas channel segments etc) (3)Processes and their interactions defined by frequency or prob-ability distributions and (4) Emergence of processes andpatterns at large scales that arise due to the collective behav-ior of large numbers of processes acting over smaller scales(Benda et al 1998) Several illustrative examples that exam-ine the effects of increasing scale in the watershed sciences aregiven below

Example One Forest Vegetation

Long-term (decadel to century) sequences of climate (rain-storms floods and fires) interacting with topography dictatepatterns of vegetation Distributions of climate topographyand vegetation are all scale dependent (eg Figures 1 and 2)

8 WMC Networker Summer 2001

In this example the scale dependency of vegetation is illus-trated using a coarse-grained fire model (Benda 1994) in thehumid temperate landscape of the Oregon Coast Range (OCR)Forest-replacing wildfires of a range of sizes (~1 ndash 1000 km2)over the last several millennia occurred at a mean interval ofabout 300 years although there was inter-millennial variabil-ity (Teensma 1987 Long 1995) In the OCR probabilitydistributions of forest ages are predicted to be positively

skewed and exponentially shaped (Benda 1994) (Figure 6)similar to fire model predictions in other landscapes (Johnsonand Van Wagener 1984) The right skewness arises because ofa decreasing probability of developing very old trees in conjunc-tion with an approximately equal susceptibility of fires acrossall age classes This forces the highest proportion of trees tooccur in the youngest age class and a gradual and systematicdecline in areas containing older trees

400 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

40 sq km

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

4000 sqkm

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

20000 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

0

005

01

015

02

025

03

0

200 400 600 800

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

035

50 250

450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650

850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

10 130

250

370

490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

012

014

016

10 130 250

370 490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

0

001

002

003

004

005

006

007

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

8 5

Figure 6 An example of scaling relationships using a forest fires and forest growth In the Oregon Coast Range fires occur onaverage once every 300 years and have sizes ranging from 1 to 1000 km2 (A) A simulated time series of mean forest age over a2500-year period showing increasing variability with decreasing spatial scale (B) Forest ages sampled in any one year will bebetter represented at larger spatial scales In addition the spatial distribution of ages at 20000 km2 should be statisticallysimilar to the long-term probability distribution of ages at any single point in the forest (C) The decreased variability of meanforest age with increasing spatial scale is represented in the form of a compressed and more symmetrical distribution ademonstration of the central limit theorem

WMC Networker Summer 2001 9

The forest fire example illustrates several of the scaling rela-tionships outlined earlier in this article Over sufficientlylarge areas the spatial distribution of forest ages measured inany year becomes similar to the expected right-skewed prob-ability distribution of ages at a single point in the forest(Figure 6) illustrating the ergodic hypothesis (Figure 5) How-ever the spatial size of the sample dictates the representationof that probability distribution illustrating a spatial scalingeffect (illustrated in Figure 2) When viewed as a populationof forest ages having mean value in each year the variabilityof the distribution of means decreases with increasing spatialscale with the distribution increasing in symmetry (Figure 6)

demonstrating the central limit theorem (Figure 3) Each ofthese patterns has geomorphological and ecological ramifica-tions In addition the increasing stability of the vegetationage distribution with increasing spatial scale has potentialimplications for characterizing environmental impacts (dis-cussed later)

Example Two Erosion by Landsliding

Erosion is dictated in large part by climate (ie rainstorms)slope steepness and vegetation Erosion often depends on athreshold being exceeded such as in rainfall-triggeredlandsliding (Caine 1990) or in sheetwash and gullying that is

km

5N

200 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

Tilton 200 km 2

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

esLandslides occur on average in 50 of every 100 years2 years out of every 100 have more than 10 landslides

km

5N

25 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10Number of Landslides

o

f Y

ears

Upper West Fork 25 km 2

Landslides occur on average in 15 of every 100 yearsOnly 3 of every 1000 year have more than 10 landslides

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

es

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

km

5N

3 km 2Upper West Fork 3 km 2

Landslides occur on average in only 4 of every 100 yearsand 3 of those 4 involve only 1 landslide

0

10

20

30

40

50

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f Lan

dsl

ides

a)

b)

c)

Figure 7The frequency of landsliding within a basin and the number of landslides occurring over any time varies systematicallywith basin size The 200 km2 area (a) encompasses thousands of potential landslide sites within this area landslides are relativelyfrequent and can occur in relatively large numbers The 25 km2 sub basin (b) contains fewer potential landslide sites landslidingwithin this area is less frequent with fewer landslides overall At 3 km2 (c) landsliding is very infrequent and any episode oflandsliding involves only a few sites

10 WMC Networker Summer 2001

dependent on soil hydrophobicity (Heede 1988) Punctuatederosion is ubiquitous in North America including in the south-ern coastal chaparral (Rice 1973) Cascade humid mountains(Swanson et al 1982) Pacific coastal rainforests (Hogan et al1995 Benda and Dunne 1997a) Appalachian Mountains (Hackand Goodlett 1960) and in the intermountain and highlandarid regions (Wohl and Pearthree 1991 Meyer et al 1995Robichaud and Brown 1999)

At the scale of populations of hillslopes over decades to centu-ries erosion occurs according to long-term patterns of climateinteracting with vegetation and topography To illustrate thiscentury-long patterns of erosion were investigated in the OCRusing a coarse-grained computational model (Benda and Dunne1997 Dunne 1998) In the simulation the temporal andspatial patterns of landsliding and debris flows are predictedto be an outcome of 1) probability distributions of stormintensity and duration 2) probability distributions of fire

recurrence and size 3) frequency distributions of topographicand geotechnical properties 4) deterministic thickening ofcolluvium in landslide sites 5) deterministic trajectories oftree rooting strength and 6) probability functions for sedi-ment transfer from hillslopes to channels The model pre-dicted that periodic fires and rainstorms triggered spates oflandsliding and debris flows every few decades to few centu-ries with little erosion at other times and places Low erosionrates punctuated by high magnitude releases of sediment atthe scale of a single hillslope or small basin is expressed by astrongly right-skewed or an exponential probability distribu-tion (Figure 7) Frequency of failures is low in small water-sheds because of the low frequency of fires and relatively smallnumber of landslide sites Landslide frequency and magni-tude increases with increasing drainage area because of in-creasing number of landslide sites and an increasing probabil-ity of storms and fires (Figure 7)

Example Three Channel Sedimentation

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

012345678910

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

00

05

10

15

0 500 1000 1500 2000 2500 3000

Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

FrequencyCumulative

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

180 km 2

45 km 2

25 km 2

00

05

10

15

20

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2) 3 km 2

Figure 8 An example of sediment fluctuations vary with spatial scale using computer simulations from southwest WashingtonThe probability distributions of fluxes evolve downstream to more symmetrical forms demonstrating the central limit theorem

WMC Networker Summer 2001 11

Over long periods channel networks obtain sediment at ratesdictated by probability distributions of erosion from indi-vidual hillslopes and small tributary basins (eg Figure 7)Sediment flux at any point in a channel network representstherefore an integrated sample of all upstream sources ofsupply Releases of sediment in a basin may be synchronousand correlated in time such as bank erosion during floods thatis likely to occur in smaller watersheds where storm size isequal to or greater than basin size Asynchronous erosion islikely the rule in basins where threshold-dependent erosionoccurs such as fire-induced surface erosion or storm triggeredmass wasting

A sediment routing model was applied to a mountain land-scape in southwest Washington to illustrate the effects ofsediment routing on the downstream distribution of sedimentflux The theoretical analysis required in addition to prob-ability distributions of erosion (eg Figure 7) probabilitydistributions of transport capacities transport velocities andparticle attrition (Benda and Dunne 1997b) Sediment fluxfrom individual hillslopes and small low-order basins is pre-

dicted to be punctaform and thereforestrongly right-skewed similar to theOCR example (Figure 7) The channelobtains sediment over long periodsfrom such numerous right-skewed dis-tributions in small headwater basinsThe accumulation of sediment fromthe multitude of sediment sources overtime causes the flux distribution toevolve into more symmetrical forms ademonstration of the central limittheorem (Figure 8 also see the moregeneral form in Figure 4) The modelalso predicts that the magnitude ofsediment perturbations decreasesdownstream in conjunction with a cor-responding increase in their numberIncrease in perturbation frequency isdue to an increasing number and there-fore probability of sediment releasesdownstream A decrease in magni-tude of the perturbation (ie differ-ence between minimum and maximumvalues) is due to several factors in-cluding 1) particle attrition (break-down) that increasingly damps sedi-ment signals with distance traveled2) a larger and persistent supply andtherefore store of gravel 3) diffusion ofdiscreet sediment pulses due to selec-tive transport and temporary storagein bars and 4) increasing channelwidth (Benda and Dunne 1997b)

Episodic introduction of sediment cancreates pulses or waves of sedimentthat migrate thought the network caus-ing changes in channel morphologyincluding variations in channel bedelevation sinuosity substrate sizespools etc (Gilbert 1917 Madej andOzake 1996 Benda 1990 Miller andBenda in press) In addition fluctua-tions in sediment supply may also

cause changes in sediment storage at a particular locationsuch as near fans or other low-gradient areas and these havebeen referred to as stationary waves (Benda and Dunne 1997b)Migrating or stationary waves create coarse-textured cut andfill terraces (Nakamura 1986 Roberts and Church 1986Miller and Benda in press) Hence the predicted fluctuationsin sediment supply by the model (Figure 8) have morphologicalconsequences to channels and valley floors and therefore on tothe biological communities that inhabit those environments

Example Four Large Woody Debris

The supply and storage of large woody debris (LWD) in streamsis governed by several landscape processes that occur punctu-ated in time over decades to centuries including by fireswindstorms landslides stand mortality and floods A woodbudgeting framework (Benda and Sias 1998 in press) wasused to evaluate how those landscape processes constrainlong-term patterns of wood abundance in streams

In the example below the effect of two end member fire

Rel

ativ

e F

requ

ency

(co

lum

ns)

Cum

ulative frequency (lines)

00

01

02

03

04

15 - 20 30 - 35 45 - 50 60 - 6500

04

06

08

150 500

150 500

Fire recurrence interval (years)

B

LWD storage (vu per 100 m reach)

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Year (first fire at year 500)

LWD

sto

rage

(vu

1

00 m

)A FIRES

(500-yr cycle)

Stand mortality

FIRES(150-yr cycle)

150500

Fire recurrence interval (years)

05 10

02

0 - 5 75 - 80

Figure 9 (A) Patterns of wood storage for fire cycles of 150 and 500 years Gradualincreases in storage represent chronic stand mortality The magnitude of the abruptpulses of wood storage is governed by the amount of standing biomass (controlled by forestage) and the time interval of the toppling of fire-killed trees (40 years in this example) (B)More frequent fires result in a compressed range of variability and a shift in thedistribution towards lower wood volumes (solid bars represent the 500-year cycle) Lessfrequent fires shift the distribution of LWD volumes towards the right into higher volumesThese patterns indicate the potential for significant differences in LWD storage alongclimatic gradients in the Pacific Northwest region (east to west and north to south)

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 7

of the population of elements Holding the mean size of adisturbance such as a fire or landslide constant the probabil-ity that the distribution of values will change over any giventime increases as the size of the population decreases (Figure3) This scaling relation is also observed in the temporalvariability of the mean value of the attribute under consider-ation This is simply a consequence that something (firelandslide flood) does not happen at the same intensity (oreven occur) every place at once The statistical stability of aspatial distribution of physical habitats (ie resilience toshifts over time) may influence such things as population sizespatial distribution and dispersal strategies

Scaling in time modulated by channel networks Scalingrelationships one through three are not only applicable tolandscapes or channel networks they are applicable to anyenvironment including galaxies or grains of sand on a beachThe fourth universal scaling relationship applies particularlyto branching networks such as a channel system Over longtime periods a channel network samples from a multitude ofindividual sources of mass flux (such as sediment and LWD)each with their own probability distribution (Figure 1) Theintegration of all sampled sources yields a long-term distribu-tion of mass flux through the network that evolves down-stream into more symmetrical forms a demonstration of thecentral limit theorem (Benda and Dunne 1997 Figure 4) Thisspatial evolution of the probability of channel and ripariandisturbances may impose spatial organization on aquatic andriparian habitats and on the species that inhabit them

Similarity Between Space and Time Variability (usingfield clues) At sufficiently large space and time scales thefrequency distribution of a spatial ensemble of elements (Fig-ure 2) may become statistically similar to a temporal distribu-tion of that attribute at a single location (Figure 1) thispotential commensurability is shown in Figure 5 For statis-tical similarity to hold all forcing functions must be tempo-rally stationary (over certain intervals) Similarity betweenlarge spatial samples and large temporal samples is referredto as an ergodic relationship (Hann 1977) and it may be usedto infer temporal behavior from a spatial sample Althoughcomplete statistical equivalence is unlikely location-for-timesubstitution (Brundsen and Thornes 1979) allows for an un-derstanding of temporal change from only a spatial sampleThis technique is particular useful for evaluating landscapebehavior including model predictions of same that encompassdecades to centuries during short human life spans Next weexamine some of the ramifications of increasing spatial andtemporal scale on watershed science management and regu-lation

Increasing Scale Implications for WatershedScienceCircumventing Limitations in Understanding CoarseGraining

Landscape interactions create patterns in the temporal be-havior and spatial distribution of watershed attributes thatmay be discernible only at large temporal and spatial scales(Figures 1 - 4) Because of the absence of long-term datacomputational models are needed to identify these patternsand the interactions that created them

Ideally large-scale computational models would be built uponphysics-based understanding of all relevant interdisciplinaryprocesses However there is a growing consensus that at leastin the near future there will be theoretical and technicalimpediments to developing large-scale environmental modelsbased on smaller-scale processes (Weinberg 1975 Dooge 1986)To circumvent present limitations watershed-scale modelscould be constructed whereby some small-scale processes areignored or are subsumed within larger-scale representationsof those processes a strategy commonly applied by physicistsand referred to as coarse graining (Gell-Mann 1994) In prac-tice in the watershed sciences this often requires combiningby means of mathematical synthesis and computer simula-tion empirical knowledge with theoretical reasoning avail-able at smaller scales to produce new understanding at largerscales (Roth et al 1989 Benda and Dunne 1997) The objec-tive of this approach would not be precise predictions aboutfuture states at individual sites but rather new testablehypotheses on large-scale interactions of climate topographyvegetation and riverine environments This approach has ahistory in the study of certain hydrological problems at largescales (Smith and Bretherton 1972 Rodriguez-Iturbe andValdez 1979)

Coarse-grained modeling is well suited to the use of distribu-tions and exploring the effects of scale on them (ie Figures 1ndash 5) System models by definition produce synthetic timeseries of watershed behavior (ie forest fires storms erosionchannel changes etc) In evaluating model predictions overhuman time frames (ie for hypothesis testing or environmen-tal problem solving) the typical absence of long-term dataforces us to consider a population of watershed elementssampled over a large spatial area Here again point observa-tions are insufficient to reveal system behavior and the distri-bution parameter in numerical models allow us to link behav-ior over large time and space scales to field observations madeover short times but large areas

Although the coarse-grained approach is taken by necessity inthe study of complex environmental systems there is consid-erable potential for obtaining new insights and understand-ing Focusing on small-scale processes often precludes seeingthe ldquobig picturerdquo that is the emergence of patterns and pro-cesses unseen at small scales Indeed the description of large-scale patterns of behavior even in the absence of mechanisticunderstanding for all of the observed processes is a hallmarkof the study of complex systems (Waldrop 1992 Kellert 1993Gell-Mann 1994) When applied to landscapes this approachyields a form of knowledge characterized by (1) Time andtherefore history (2) Populations of landscape elements (egforest stands erosion source areas channel segments etc) (3)Processes and their interactions defined by frequency or prob-ability distributions and (4) Emergence of processes andpatterns at large scales that arise due to the collective behav-ior of large numbers of processes acting over smaller scales(Benda et al 1998) Several illustrative examples that exam-ine the effects of increasing scale in the watershed sciences aregiven below

Example One Forest Vegetation

Long-term (decadel to century) sequences of climate (rain-storms floods and fires) interacting with topography dictatepatterns of vegetation Distributions of climate topographyand vegetation are all scale dependent (eg Figures 1 and 2)

8 WMC Networker Summer 2001

In this example the scale dependency of vegetation is illus-trated using a coarse-grained fire model (Benda 1994) in thehumid temperate landscape of the Oregon Coast Range (OCR)Forest-replacing wildfires of a range of sizes (~1 ndash 1000 km2)over the last several millennia occurred at a mean interval ofabout 300 years although there was inter-millennial variabil-ity (Teensma 1987 Long 1995) In the OCR probabilitydistributions of forest ages are predicted to be positively

skewed and exponentially shaped (Benda 1994) (Figure 6)similar to fire model predictions in other landscapes (Johnsonand Van Wagener 1984) The right skewness arises because ofa decreasing probability of developing very old trees in conjunc-tion with an approximately equal susceptibility of fires acrossall age classes This forces the highest proportion of trees tooccur in the youngest age class and a gradual and systematicdecline in areas containing older trees

400 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

40 sq km

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

4000 sqkm

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

20000 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

0

005

01

015

02

025

03

0

200 400 600 800

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

035

50 250

450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650

850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

10 130

250

370

490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

012

014

016

10 130 250

370 490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

0

001

002

003

004

005

006

007

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

8 5

Figure 6 An example of scaling relationships using a forest fires and forest growth In the Oregon Coast Range fires occur onaverage once every 300 years and have sizes ranging from 1 to 1000 km2 (A) A simulated time series of mean forest age over a2500-year period showing increasing variability with decreasing spatial scale (B) Forest ages sampled in any one year will bebetter represented at larger spatial scales In addition the spatial distribution of ages at 20000 km2 should be statisticallysimilar to the long-term probability distribution of ages at any single point in the forest (C) The decreased variability of meanforest age with increasing spatial scale is represented in the form of a compressed and more symmetrical distribution ademonstration of the central limit theorem

WMC Networker Summer 2001 9

The forest fire example illustrates several of the scaling rela-tionships outlined earlier in this article Over sufficientlylarge areas the spatial distribution of forest ages measured inany year becomes similar to the expected right-skewed prob-ability distribution of ages at a single point in the forest(Figure 6) illustrating the ergodic hypothesis (Figure 5) How-ever the spatial size of the sample dictates the representationof that probability distribution illustrating a spatial scalingeffect (illustrated in Figure 2) When viewed as a populationof forest ages having mean value in each year the variabilityof the distribution of means decreases with increasing spatialscale with the distribution increasing in symmetry (Figure 6)

demonstrating the central limit theorem (Figure 3) Each ofthese patterns has geomorphological and ecological ramifica-tions In addition the increasing stability of the vegetationage distribution with increasing spatial scale has potentialimplications for characterizing environmental impacts (dis-cussed later)

Example Two Erosion by Landsliding

Erosion is dictated in large part by climate (ie rainstorms)slope steepness and vegetation Erosion often depends on athreshold being exceeded such as in rainfall-triggeredlandsliding (Caine 1990) or in sheetwash and gullying that is

km

5N

200 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

Tilton 200 km 2

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

esLandslides occur on average in 50 of every 100 years2 years out of every 100 have more than 10 landslides

km

5N

25 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10Number of Landslides

o

f Y

ears

Upper West Fork 25 km 2

Landslides occur on average in 15 of every 100 yearsOnly 3 of every 1000 year have more than 10 landslides

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

es

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

km

5N

3 km 2Upper West Fork 3 km 2

Landslides occur on average in only 4 of every 100 yearsand 3 of those 4 involve only 1 landslide

0

10

20

30

40

50

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f Lan

dsl

ides

a)

b)

c)

Figure 7The frequency of landsliding within a basin and the number of landslides occurring over any time varies systematicallywith basin size The 200 km2 area (a) encompasses thousands of potential landslide sites within this area landslides are relativelyfrequent and can occur in relatively large numbers The 25 km2 sub basin (b) contains fewer potential landslide sites landslidingwithin this area is less frequent with fewer landslides overall At 3 km2 (c) landsliding is very infrequent and any episode oflandsliding involves only a few sites

10 WMC Networker Summer 2001

dependent on soil hydrophobicity (Heede 1988) Punctuatederosion is ubiquitous in North America including in the south-ern coastal chaparral (Rice 1973) Cascade humid mountains(Swanson et al 1982) Pacific coastal rainforests (Hogan et al1995 Benda and Dunne 1997a) Appalachian Mountains (Hackand Goodlett 1960) and in the intermountain and highlandarid regions (Wohl and Pearthree 1991 Meyer et al 1995Robichaud and Brown 1999)

At the scale of populations of hillslopes over decades to centu-ries erosion occurs according to long-term patterns of climateinteracting with vegetation and topography To illustrate thiscentury-long patterns of erosion were investigated in the OCRusing a coarse-grained computational model (Benda and Dunne1997 Dunne 1998) In the simulation the temporal andspatial patterns of landsliding and debris flows are predictedto be an outcome of 1) probability distributions of stormintensity and duration 2) probability distributions of fire

recurrence and size 3) frequency distributions of topographicand geotechnical properties 4) deterministic thickening ofcolluvium in landslide sites 5) deterministic trajectories oftree rooting strength and 6) probability functions for sedi-ment transfer from hillslopes to channels The model pre-dicted that periodic fires and rainstorms triggered spates oflandsliding and debris flows every few decades to few centu-ries with little erosion at other times and places Low erosionrates punctuated by high magnitude releases of sediment atthe scale of a single hillslope or small basin is expressed by astrongly right-skewed or an exponential probability distribu-tion (Figure 7) Frequency of failures is low in small water-sheds because of the low frequency of fires and relatively smallnumber of landslide sites Landslide frequency and magni-tude increases with increasing drainage area because of in-creasing number of landslide sites and an increasing probabil-ity of storms and fires (Figure 7)

Example Three Channel Sedimentation

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

012345678910

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

00

05

10

15

0 500 1000 1500 2000 2500 3000

Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

FrequencyCumulative

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

180 km 2

45 km 2

25 km 2

00

05

10

15

20

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2) 3 km 2

Figure 8 An example of sediment fluctuations vary with spatial scale using computer simulations from southwest WashingtonThe probability distributions of fluxes evolve downstream to more symmetrical forms demonstrating the central limit theorem

WMC Networker Summer 2001 11

Over long periods channel networks obtain sediment at ratesdictated by probability distributions of erosion from indi-vidual hillslopes and small tributary basins (eg Figure 7)Sediment flux at any point in a channel network representstherefore an integrated sample of all upstream sources ofsupply Releases of sediment in a basin may be synchronousand correlated in time such as bank erosion during floods thatis likely to occur in smaller watersheds where storm size isequal to or greater than basin size Asynchronous erosion islikely the rule in basins where threshold-dependent erosionoccurs such as fire-induced surface erosion or storm triggeredmass wasting

A sediment routing model was applied to a mountain land-scape in southwest Washington to illustrate the effects ofsediment routing on the downstream distribution of sedimentflux The theoretical analysis required in addition to prob-ability distributions of erosion (eg Figure 7) probabilitydistributions of transport capacities transport velocities andparticle attrition (Benda and Dunne 1997b) Sediment fluxfrom individual hillslopes and small low-order basins is pre-

dicted to be punctaform and thereforestrongly right-skewed similar to theOCR example (Figure 7) The channelobtains sediment over long periodsfrom such numerous right-skewed dis-tributions in small headwater basinsThe accumulation of sediment fromthe multitude of sediment sources overtime causes the flux distribution toevolve into more symmetrical forms ademonstration of the central limittheorem (Figure 8 also see the moregeneral form in Figure 4) The modelalso predicts that the magnitude ofsediment perturbations decreasesdownstream in conjunction with a cor-responding increase in their numberIncrease in perturbation frequency isdue to an increasing number and there-fore probability of sediment releasesdownstream A decrease in magni-tude of the perturbation (ie differ-ence between minimum and maximumvalues) is due to several factors in-cluding 1) particle attrition (break-down) that increasingly damps sedi-ment signals with distance traveled2) a larger and persistent supply andtherefore store of gravel 3) diffusion ofdiscreet sediment pulses due to selec-tive transport and temporary storagein bars and 4) increasing channelwidth (Benda and Dunne 1997b)

Episodic introduction of sediment cancreates pulses or waves of sedimentthat migrate thought the network caus-ing changes in channel morphologyincluding variations in channel bedelevation sinuosity substrate sizespools etc (Gilbert 1917 Madej andOzake 1996 Benda 1990 Miller andBenda in press) In addition fluctua-tions in sediment supply may also

cause changes in sediment storage at a particular locationsuch as near fans or other low-gradient areas and these havebeen referred to as stationary waves (Benda and Dunne 1997b)Migrating or stationary waves create coarse-textured cut andfill terraces (Nakamura 1986 Roberts and Church 1986Miller and Benda in press) Hence the predicted fluctuationsin sediment supply by the model (Figure 8) have morphologicalconsequences to channels and valley floors and therefore on tothe biological communities that inhabit those environments

Example Four Large Woody Debris

The supply and storage of large woody debris (LWD) in streamsis governed by several landscape processes that occur punctu-ated in time over decades to centuries including by fireswindstorms landslides stand mortality and floods A woodbudgeting framework (Benda and Sias 1998 in press) wasused to evaluate how those landscape processes constrainlong-term patterns of wood abundance in streams

In the example below the effect of two end member fire

Rel

ativ

e F

requ

ency

(co

lum

ns)

Cum

ulative frequency (lines)

00

01

02

03

04

15 - 20 30 - 35 45 - 50 60 - 6500

04

06

08

150 500

150 500

Fire recurrence interval (years)

B

LWD storage (vu per 100 m reach)

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Year (first fire at year 500)

LWD

sto

rage

(vu

1

00 m

)A FIRES

(500-yr cycle)

Stand mortality

FIRES(150-yr cycle)

150500

Fire recurrence interval (years)

05 10

02

0 - 5 75 - 80

Figure 9 (A) Patterns of wood storage for fire cycles of 150 and 500 years Gradualincreases in storage represent chronic stand mortality The magnitude of the abruptpulses of wood storage is governed by the amount of standing biomass (controlled by forestage) and the time interval of the toppling of fire-killed trees (40 years in this example) (B)More frequent fires result in a compressed range of variability and a shift in thedistribution towards lower wood volumes (solid bars represent the 500-year cycle) Lessfrequent fires shift the distribution of LWD volumes towards the right into higher volumesThese patterns indicate the potential for significant differences in LWD storage alongclimatic gradients in the Pacific Northwest region (east to west and north to south)

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

8 WMC Networker Summer 2001

In this example the scale dependency of vegetation is illus-trated using a coarse-grained fire model (Benda 1994) in thehumid temperate landscape of the Oregon Coast Range (OCR)Forest-replacing wildfires of a range of sizes (~1 ndash 1000 km2)over the last several millennia occurred at a mean interval ofabout 300 years although there was inter-millennial variabil-ity (Teensma 1987 Long 1995) In the OCR probabilitydistributions of forest ages are predicted to be positively

skewed and exponentially shaped (Benda 1994) (Figure 6)similar to fire model predictions in other landscapes (Johnsonand Van Wagener 1984) The right skewness arises because ofa decreasing probability of developing very old trees in conjunc-tion with an approximately equal susceptibility of fires acrossall age classes This forces the highest proportion of trees tooccur in the youngest age class and a gradual and systematicdecline in areas containing older trees

400 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

40 sq km

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

4000 sqkm

0

500

0 1000 2000 3000

TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

20000 sq km

0

200

400

600

0 1000 2000 3000TIME (years)

ME

AN

FO

RE

ST

AG

E (

year

s)

0

005

01

015

02

025

03

0

200 400 600 800

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

035

50 250

450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650

850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

50 250 450 650 850

FOREST AGE (YRS)

PE

RC

EN

T

0

005

01

015

02

025

03

10 130

250

370

490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

012

014

016

10 130 250

370 490

MEAN AGE

PE

RC

EN

T

0

002

004

006

008

01

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

0

001

002

003

004

005

006

007

10 120

230

340

450

560

MEAN AGE

PE

RC

EN

T

8 5

Figure 6 An example of scaling relationships using a forest fires and forest growth In the Oregon Coast Range fires occur onaverage once every 300 years and have sizes ranging from 1 to 1000 km2 (A) A simulated time series of mean forest age over a2500-year period showing increasing variability with decreasing spatial scale (B) Forest ages sampled in any one year will bebetter represented at larger spatial scales In addition the spatial distribution of ages at 20000 km2 should be statisticallysimilar to the long-term probability distribution of ages at any single point in the forest (C) The decreased variability of meanforest age with increasing spatial scale is represented in the form of a compressed and more symmetrical distribution ademonstration of the central limit theorem

WMC Networker Summer 2001 9

The forest fire example illustrates several of the scaling rela-tionships outlined earlier in this article Over sufficientlylarge areas the spatial distribution of forest ages measured inany year becomes similar to the expected right-skewed prob-ability distribution of ages at a single point in the forest(Figure 6) illustrating the ergodic hypothesis (Figure 5) How-ever the spatial size of the sample dictates the representationof that probability distribution illustrating a spatial scalingeffect (illustrated in Figure 2) When viewed as a populationof forest ages having mean value in each year the variabilityof the distribution of means decreases with increasing spatialscale with the distribution increasing in symmetry (Figure 6)

demonstrating the central limit theorem (Figure 3) Each ofthese patterns has geomorphological and ecological ramifica-tions In addition the increasing stability of the vegetationage distribution with increasing spatial scale has potentialimplications for characterizing environmental impacts (dis-cussed later)

Example Two Erosion by Landsliding

Erosion is dictated in large part by climate (ie rainstorms)slope steepness and vegetation Erosion often depends on athreshold being exceeded such as in rainfall-triggeredlandsliding (Caine 1990) or in sheetwash and gullying that is

km

5N

200 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

Tilton 200 km 2

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

esLandslides occur on average in 50 of every 100 years2 years out of every 100 have more than 10 landslides

km

5N

25 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10Number of Landslides

o

f Y

ears

Upper West Fork 25 km 2

Landslides occur on average in 15 of every 100 yearsOnly 3 of every 1000 year have more than 10 landslides

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

es

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

km

5N

3 km 2Upper West Fork 3 km 2

Landslides occur on average in only 4 of every 100 yearsand 3 of those 4 involve only 1 landslide

0

10

20

30

40

50

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f Lan

dsl

ides

a)

b)

c)

Figure 7The frequency of landsliding within a basin and the number of landslides occurring over any time varies systematicallywith basin size The 200 km2 area (a) encompasses thousands of potential landslide sites within this area landslides are relativelyfrequent and can occur in relatively large numbers The 25 km2 sub basin (b) contains fewer potential landslide sites landslidingwithin this area is less frequent with fewer landslides overall At 3 km2 (c) landsliding is very infrequent and any episode oflandsliding involves only a few sites

10 WMC Networker Summer 2001

dependent on soil hydrophobicity (Heede 1988) Punctuatederosion is ubiquitous in North America including in the south-ern coastal chaparral (Rice 1973) Cascade humid mountains(Swanson et al 1982) Pacific coastal rainforests (Hogan et al1995 Benda and Dunne 1997a) Appalachian Mountains (Hackand Goodlett 1960) and in the intermountain and highlandarid regions (Wohl and Pearthree 1991 Meyer et al 1995Robichaud and Brown 1999)

At the scale of populations of hillslopes over decades to centu-ries erosion occurs according to long-term patterns of climateinteracting with vegetation and topography To illustrate thiscentury-long patterns of erosion were investigated in the OCRusing a coarse-grained computational model (Benda and Dunne1997 Dunne 1998) In the simulation the temporal andspatial patterns of landsliding and debris flows are predictedto be an outcome of 1) probability distributions of stormintensity and duration 2) probability distributions of fire

recurrence and size 3) frequency distributions of topographicand geotechnical properties 4) deterministic thickening ofcolluvium in landslide sites 5) deterministic trajectories oftree rooting strength and 6) probability functions for sedi-ment transfer from hillslopes to channels The model pre-dicted that periodic fires and rainstorms triggered spates oflandsliding and debris flows every few decades to few centu-ries with little erosion at other times and places Low erosionrates punctuated by high magnitude releases of sediment atthe scale of a single hillslope or small basin is expressed by astrongly right-skewed or an exponential probability distribu-tion (Figure 7) Frequency of failures is low in small water-sheds because of the low frequency of fires and relatively smallnumber of landslide sites Landslide frequency and magni-tude increases with increasing drainage area because of in-creasing number of landslide sites and an increasing probabil-ity of storms and fires (Figure 7)

Example Three Channel Sedimentation

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

012345678910

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

00

05

10

15

0 500 1000 1500 2000 2500 3000

Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

FrequencyCumulative

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

180 km 2

45 km 2

25 km 2

00

05

10

15

20

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2) 3 km 2

Figure 8 An example of sediment fluctuations vary with spatial scale using computer simulations from southwest WashingtonThe probability distributions of fluxes evolve downstream to more symmetrical forms demonstrating the central limit theorem

WMC Networker Summer 2001 11

Over long periods channel networks obtain sediment at ratesdictated by probability distributions of erosion from indi-vidual hillslopes and small tributary basins (eg Figure 7)Sediment flux at any point in a channel network representstherefore an integrated sample of all upstream sources ofsupply Releases of sediment in a basin may be synchronousand correlated in time such as bank erosion during floods thatis likely to occur in smaller watersheds where storm size isequal to or greater than basin size Asynchronous erosion islikely the rule in basins where threshold-dependent erosionoccurs such as fire-induced surface erosion or storm triggeredmass wasting

A sediment routing model was applied to a mountain land-scape in southwest Washington to illustrate the effects ofsediment routing on the downstream distribution of sedimentflux The theoretical analysis required in addition to prob-ability distributions of erosion (eg Figure 7) probabilitydistributions of transport capacities transport velocities andparticle attrition (Benda and Dunne 1997b) Sediment fluxfrom individual hillslopes and small low-order basins is pre-

dicted to be punctaform and thereforestrongly right-skewed similar to theOCR example (Figure 7) The channelobtains sediment over long periodsfrom such numerous right-skewed dis-tributions in small headwater basinsThe accumulation of sediment fromthe multitude of sediment sources overtime causes the flux distribution toevolve into more symmetrical forms ademonstration of the central limittheorem (Figure 8 also see the moregeneral form in Figure 4) The modelalso predicts that the magnitude ofsediment perturbations decreasesdownstream in conjunction with a cor-responding increase in their numberIncrease in perturbation frequency isdue to an increasing number and there-fore probability of sediment releasesdownstream A decrease in magni-tude of the perturbation (ie differ-ence between minimum and maximumvalues) is due to several factors in-cluding 1) particle attrition (break-down) that increasingly damps sedi-ment signals with distance traveled2) a larger and persistent supply andtherefore store of gravel 3) diffusion ofdiscreet sediment pulses due to selec-tive transport and temporary storagein bars and 4) increasing channelwidth (Benda and Dunne 1997b)

Episodic introduction of sediment cancreates pulses or waves of sedimentthat migrate thought the network caus-ing changes in channel morphologyincluding variations in channel bedelevation sinuosity substrate sizespools etc (Gilbert 1917 Madej andOzake 1996 Benda 1990 Miller andBenda in press) In addition fluctua-tions in sediment supply may also

cause changes in sediment storage at a particular locationsuch as near fans or other low-gradient areas and these havebeen referred to as stationary waves (Benda and Dunne 1997b)Migrating or stationary waves create coarse-textured cut andfill terraces (Nakamura 1986 Roberts and Church 1986Miller and Benda in press) Hence the predicted fluctuationsin sediment supply by the model (Figure 8) have morphologicalconsequences to channels and valley floors and therefore on tothe biological communities that inhabit those environments

Example Four Large Woody Debris

The supply and storage of large woody debris (LWD) in streamsis governed by several landscape processes that occur punctu-ated in time over decades to centuries including by fireswindstorms landslides stand mortality and floods A woodbudgeting framework (Benda and Sias 1998 in press) wasused to evaluate how those landscape processes constrainlong-term patterns of wood abundance in streams

In the example below the effect of two end member fire

Rel

ativ

e F

requ

ency

(co

lum

ns)

Cum

ulative frequency (lines)

00

01

02

03

04

15 - 20 30 - 35 45 - 50 60 - 6500

04

06

08

150 500

150 500

Fire recurrence interval (years)

B

LWD storage (vu per 100 m reach)

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Year (first fire at year 500)

LWD

sto

rage

(vu

1

00 m

)A FIRES

(500-yr cycle)

Stand mortality

FIRES(150-yr cycle)

150500

Fire recurrence interval (years)

05 10

02

0 - 5 75 - 80

Figure 9 (A) Patterns of wood storage for fire cycles of 150 and 500 years Gradualincreases in storage represent chronic stand mortality The magnitude of the abruptpulses of wood storage is governed by the amount of standing biomass (controlled by forestage) and the time interval of the toppling of fire-killed trees (40 years in this example) (B)More frequent fires result in a compressed range of variability and a shift in thedistribution towards lower wood volumes (solid bars represent the 500-year cycle) Lessfrequent fires shift the distribution of LWD volumes towards the right into higher volumesThese patterns indicate the potential for significant differences in LWD storage alongclimatic gradients in the Pacific Northwest region (east to west and north to south)

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 9

The forest fire example illustrates several of the scaling rela-tionships outlined earlier in this article Over sufficientlylarge areas the spatial distribution of forest ages measured inany year becomes similar to the expected right-skewed prob-ability distribution of ages at a single point in the forest(Figure 6) illustrating the ergodic hypothesis (Figure 5) How-ever the spatial size of the sample dictates the representationof that probability distribution illustrating a spatial scalingeffect (illustrated in Figure 2) When viewed as a populationof forest ages having mean value in each year the variabilityof the distribution of means decreases with increasing spatialscale with the distribution increasing in symmetry (Figure 6)

demonstrating the central limit theorem (Figure 3) Each ofthese patterns has geomorphological and ecological ramifica-tions In addition the increasing stability of the vegetationage distribution with increasing spatial scale has potentialimplications for characterizing environmental impacts (dis-cussed later)

Example Two Erosion by Landsliding

Erosion is dictated in large part by climate (ie rainstorms)slope steepness and vegetation Erosion often depends on athreshold being exceeded such as in rainfall-triggeredlandsliding (Caine 1990) or in sheetwash and gullying that is

km

5N

200 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

Tilton 200 km 2

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

esLandslides occur on average in 50 of every 100 years2 years out of every 100 have more than 10 landslides

km

5N

25 km 2

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10Number of Landslides

o

f Y

ears

Upper West Fork 25 km 2

Landslides occur on average in 15 of every 100 yearsOnly 3 of every 1000 year have more than 10 landslides

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f L

and

slid

es

0

5

10

15

20

25

1 2 3 4 5 6 7 8 9 10

Number of Landslides

o

f Y

ears

km

5N

3 km 2Upper West Fork 3 km 2

Landslides occur on average in only 4 of every 100 yearsand 3 of those 4 involve only 1 landslide

0

10

20

30

40

50

0 1000 2000 3000 4000 5000Simulation Year

Nu

mb

er o

f Lan

dsl

ides

a)

b)

c)

Figure 7The frequency of landsliding within a basin and the number of landslides occurring over any time varies systematicallywith basin size The 200 km2 area (a) encompasses thousands of potential landslide sites within this area landslides are relativelyfrequent and can occur in relatively large numbers The 25 km2 sub basin (b) contains fewer potential landslide sites landslidingwithin this area is less frequent with fewer landslides overall At 3 km2 (c) landsliding is very infrequent and any episode oflandsliding involves only a few sites

10 WMC Networker Summer 2001

dependent on soil hydrophobicity (Heede 1988) Punctuatederosion is ubiquitous in North America including in the south-ern coastal chaparral (Rice 1973) Cascade humid mountains(Swanson et al 1982) Pacific coastal rainforests (Hogan et al1995 Benda and Dunne 1997a) Appalachian Mountains (Hackand Goodlett 1960) and in the intermountain and highlandarid regions (Wohl and Pearthree 1991 Meyer et al 1995Robichaud and Brown 1999)

At the scale of populations of hillslopes over decades to centu-ries erosion occurs according to long-term patterns of climateinteracting with vegetation and topography To illustrate thiscentury-long patterns of erosion were investigated in the OCRusing a coarse-grained computational model (Benda and Dunne1997 Dunne 1998) In the simulation the temporal andspatial patterns of landsliding and debris flows are predictedto be an outcome of 1) probability distributions of stormintensity and duration 2) probability distributions of fire

recurrence and size 3) frequency distributions of topographicand geotechnical properties 4) deterministic thickening ofcolluvium in landslide sites 5) deterministic trajectories oftree rooting strength and 6) probability functions for sedi-ment transfer from hillslopes to channels The model pre-dicted that periodic fires and rainstorms triggered spates oflandsliding and debris flows every few decades to few centu-ries with little erosion at other times and places Low erosionrates punctuated by high magnitude releases of sediment atthe scale of a single hillslope or small basin is expressed by astrongly right-skewed or an exponential probability distribu-tion (Figure 7) Frequency of failures is low in small water-sheds because of the low frequency of fires and relatively smallnumber of landslide sites Landslide frequency and magni-tude increases with increasing drainage area because of in-creasing number of landslide sites and an increasing probabil-ity of storms and fires (Figure 7)

Example Three Channel Sedimentation

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

012345678910

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

00

05

10

15

0 500 1000 1500 2000 2500 3000

Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

FrequencyCumulative

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

180 km 2

45 km 2

25 km 2

00

05

10

15

20

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2) 3 km 2

Figure 8 An example of sediment fluctuations vary with spatial scale using computer simulations from southwest WashingtonThe probability distributions of fluxes evolve downstream to more symmetrical forms demonstrating the central limit theorem

WMC Networker Summer 2001 11

Over long periods channel networks obtain sediment at ratesdictated by probability distributions of erosion from indi-vidual hillslopes and small tributary basins (eg Figure 7)Sediment flux at any point in a channel network representstherefore an integrated sample of all upstream sources ofsupply Releases of sediment in a basin may be synchronousand correlated in time such as bank erosion during floods thatis likely to occur in smaller watersheds where storm size isequal to or greater than basin size Asynchronous erosion islikely the rule in basins where threshold-dependent erosionoccurs such as fire-induced surface erosion or storm triggeredmass wasting

A sediment routing model was applied to a mountain land-scape in southwest Washington to illustrate the effects ofsediment routing on the downstream distribution of sedimentflux The theoretical analysis required in addition to prob-ability distributions of erosion (eg Figure 7) probabilitydistributions of transport capacities transport velocities andparticle attrition (Benda and Dunne 1997b) Sediment fluxfrom individual hillslopes and small low-order basins is pre-

dicted to be punctaform and thereforestrongly right-skewed similar to theOCR example (Figure 7) The channelobtains sediment over long periodsfrom such numerous right-skewed dis-tributions in small headwater basinsThe accumulation of sediment fromthe multitude of sediment sources overtime causes the flux distribution toevolve into more symmetrical forms ademonstration of the central limittheorem (Figure 8 also see the moregeneral form in Figure 4) The modelalso predicts that the magnitude ofsediment perturbations decreasesdownstream in conjunction with a cor-responding increase in their numberIncrease in perturbation frequency isdue to an increasing number and there-fore probability of sediment releasesdownstream A decrease in magni-tude of the perturbation (ie differ-ence between minimum and maximumvalues) is due to several factors in-cluding 1) particle attrition (break-down) that increasingly damps sedi-ment signals with distance traveled2) a larger and persistent supply andtherefore store of gravel 3) diffusion ofdiscreet sediment pulses due to selec-tive transport and temporary storagein bars and 4) increasing channelwidth (Benda and Dunne 1997b)

Episodic introduction of sediment cancreates pulses or waves of sedimentthat migrate thought the network caus-ing changes in channel morphologyincluding variations in channel bedelevation sinuosity substrate sizespools etc (Gilbert 1917 Madej andOzake 1996 Benda 1990 Miller andBenda in press) In addition fluctua-tions in sediment supply may also

cause changes in sediment storage at a particular locationsuch as near fans or other low-gradient areas and these havebeen referred to as stationary waves (Benda and Dunne 1997b)Migrating or stationary waves create coarse-textured cut andfill terraces (Nakamura 1986 Roberts and Church 1986Miller and Benda in press) Hence the predicted fluctuationsin sediment supply by the model (Figure 8) have morphologicalconsequences to channels and valley floors and therefore on tothe biological communities that inhabit those environments

Example Four Large Woody Debris

The supply and storage of large woody debris (LWD) in streamsis governed by several landscape processes that occur punctu-ated in time over decades to centuries including by fireswindstorms landslides stand mortality and floods A woodbudgeting framework (Benda and Sias 1998 in press) wasused to evaluate how those landscape processes constrainlong-term patterns of wood abundance in streams

In the example below the effect of two end member fire

Rel

ativ

e F

requ

ency

(co

lum

ns)

Cum

ulative frequency (lines)

00

01

02

03

04

15 - 20 30 - 35 45 - 50 60 - 6500

04

06

08

150 500

150 500

Fire recurrence interval (years)

B

LWD storage (vu per 100 m reach)

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Year (first fire at year 500)

LWD

sto

rage

(vu

1

00 m

)A FIRES

(500-yr cycle)

Stand mortality

FIRES(150-yr cycle)

150500

Fire recurrence interval (years)

05 10

02

0 - 5 75 - 80

Figure 9 (A) Patterns of wood storage for fire cycles of 150 and 500 years Gradualincreases in storage represent chronic stand mortality The magnitude of the abruptpulses of wood storage is governed by the amount of standing biomass (controlled by forestage) and the time interval of the toppling of fire-killed trees (40 years in this example) (B)More frequent fires result in a compressed range of variability and a shift in thedistribution towards lower wood volumes (solid bars represent the 500-year cycle) Lessfrequent fires shift the distribution of LWD volumes towards the right into higher volumesThese patterns indicate the potential for significant differences in LWD storage alongclimatic gradients in the Pacific Northwest region (east to west and north to south)

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

10 WMC Networker Summer 2001

dependent on soil hydrophobicity (Heede 1988) Punctuatederosion is ubiquitous in North America including in the south-ern coastal chaparral (Rice 1973) Cascade humid mountains(Swanson et al 1982) Pacific coastal rainforests (Hogan et al1995 Benda and Dunne 1997a) Appalachian Mountains (Hackand Goodlett 1960) and in the intermountain and highlandarid regions (Wohl and Pearthree 1991 Meyer et al 1995Robichaud and Brown 1999)

At the scale of populations of hillslopes over decades to centu-ries erosion occurs according to long-term patterns of climateinteracting with vegetation and topography To illustrate thiscentury-long patterns of erosion were investigated in the OCRusing a coarse-grained computational model (Benda and Dunne1997 Dunne 1998) In the simulation the temporal andspatial patterns of landsliding and debris flows are predictedto be an outcome of 1) probability distributions of stormintensity and duration 2) probability distributions of fire

recurrence and size 3) frequency distributions of topographicand geotechnical properties 4) deterministic thickening ofcolluvium in landslide sites 5) deterministic trajectories oftree rooting strength and 6) probability functions for sedi-ment transfer from hillslopes to channels The model pre-dicted that periodic fires and rainstorms triggered spates oflandsliding and debris flows every few decades to few centu-ries with little erosion at other times and places Low erosionrates punctuated by high magnitude releases of sediment atthe scale of a single hillslope or small basin is expressed by astrongly right-skewed or an exponential probability distribu-tion (Figure 7) Frequency of failures is low in small water-sheds because of the low frequency of fires and relatively smallnumber of landslide sites Landslide frequency and magni-tude increases with increasing drainage area because of in-creasing number of landslide sites and an increasing probabil-ity of storms and fires (Figure 7)

Example Three Channel Sedimentation

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

012345678910

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

00

05

10

15

0 500 1000 1500 2000 2500 3000

Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

FrequencyCumulative

00

05

10

15

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2)

0102030405060708090

100

0 02 04 06 08 1Volume (m3m2)

0123456789

10

180 km 2

45 km 2

25 km 2

00

05

10

15

20

0 500 1000 1500 2000 2500 3000Year

Vo

lum

e (m

3m

2) 3 km 2

Figure 8 An example of sediment fluctuations vary with spatial scale using computer simulations from southwest WashingtonThe probability distributions of fluxes evolve downstream to more symmetrical forms demonstrating the central limit theorem

WMC Networker Summer 2001 11

Over long periods channel networks obtain sediment at ratesdictated by probability distributions of erosion from indi-vidual hillslopes and small tributary basins (eg Figure 7)Sediment flux at any point in a channel network representstherefore an integrated sample of all upstream sources ofsupply Releases of sediment in a basin may be synchronousand correlated in time such as bank erosion during floods thatis likely to occur in smaller watersheds where storm size isequal to or greater than basin size Asynchronous erosion islikely the rule in basins where threshold-dependent erosionoccurs such as fire-induced surface erosion or storm triggeredmass wasting

A sediment routing model was applied to a mountain land-scape in southwest Washington to illustrate the effects ofsediment routing on the downstream distribution of sedimentflux The theoretical analysis required in addition to prob-ability distributions of erosion (eg Figure 7) probabilitydistributions of transport capacities transport velocities andparticle attrition (Benda and Dunne 1997b) Sediment fluxfrom individual hillslopes and small low-order basins is pre-

dicted to be punctaform and thereforestrongly right-skewed similar to theOCR example (Figure 7) The channelobtains sediment over long periodsfrom such numerous right-skewed dis-tributions in small headwater basinsThe accumulation of sediment fromthe multitude of sediment sources overtime causes the flux distribution toevolve into more symmetrical forms ademonstration of the central limittheorem (Figure 8 also see the moregeneral form in Figure 4) The modelalso predicts that the magnitude ofsediment perturbations decreasesdownstream in conjunction with a cor-responding increase in their numberIncrease in perturbation frequency isdue to an increasing number and there-fore probability of sediment releasesdownstream A decrease in magni-tude of the perturbation (ie differ-ence between minimum and maximumvalues) is due to several factors in-cluding 1) particle attrition (break-down) that increasingly damps sedi-ment signals with distance traveled2) a larger and persistent supply andtherefore store of gravel 3) diffusion ofdiscreet sediment pulses due to selec-tive transport and temporary storagein bars and 4) increasing channelwidth (Benda and Dunne 1997b)

Episodic introduction of sediment cancreates pulses or waves of sedimentthat migrate thought the network caus-ing changes in channel morphologyincluding variations in channel bedelevation sinuosity substrate sizespools etc (Gilbert 1917 Madej andOzake 1996 Benda 1990 Miller andBenda in press) In addition fluctua-tions in sediment supply may also

cause changes in sediment storage at a particular locationsuch as near fans or other low-gradient areas and these havebeen referred to as stationary waves (Benda and Dunne 1997b)Migrating or stationary waves create coarse-textured cut andfill terraces (Nakamura 1986 Roberts and Church 1986Miller and Benda in press) Hence the predicted fluctuationsin sediment supply by the model (Figure 8) have morphologicalconsequences to channels and valley floors and therefore on tothe biological communities that inhabit those environments

Example Four Large Woody Debris

The supply and storage of large woody debris (LWD) in streamsis governed by several landscape processes that occur punctu-ated in time over decades to centuries including by fireswindstorms landslides stand mortality and floods A woodbudgeting framework (Benda and Sias 1998 in press) wasused to evaluate how those landscape processes constrainlong-term patterns of wood abundance in streams

In the example below the effect of two end member fire

Rel

ativ

e F

requ

ency

(co

lum

ns)

Cum

ulative frequency (lines)

00

01

02

03

04

15 - 20 30 - 35 45 - 50 60 - 6500

04

06

08

150 500

150 500

Fire recurrence interval (years)

B

LWD storage (vu per 100 m reach)

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Year (first fire at year 500)

LWD

sto

rage

(vu

1

00 m

)A FIRES

(500-yr cycle)

Stand mortality

FIRES(150-yr cycle)

150500

Fire recurrence interval (years)

05 10

02

0 - 5 75 - 80

Figure 9 (A) Patterns of wood storage for fire cycles of 150 and 500 years Gradualincreases in storage represent chronic stand mortality The magnitude of the abruptpulses of wood storage is governed by the amount of standing biomass (controlled by forestage) and the time interval of the toppling of fire-killed trees (40 years in this example) (B)More frequent fires result in a compressed range of variability and a shift in thedistribution towards lower wood volumes (solid bars represent the 500-year cycle) Lessfrequent fires shift the distribution of LWD volumes towards the right into higher volumesThese patterns indicate the potential for significant differences in LWD storage alongclimatic gradients in the Pacific Northwest region (east to west and north to south)

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 11

Over long periods channel networks obtain sediment at ratesdictated by probability distributions of erosion from indi-vidual hillslopes and small tributary basins (eg Figure 7)Sediment flux at any point in a channel network representstherefore an integrated sample of all upstream sources ofsupply Releases of sediment in a basin may be synchronousand correlated in time such as bank erosion during floods thatis likely to occur in smaller watersheds where storm size isequal to or greater than basin size Asynchronous erosion islikely the rule in basins where threshold-dependent erosionoccurs such as fire-induced surface erosion or storm triggeredmass wasting

A sediment routing model was applied to a mountain land-scape in southwest Washington to illustrate the effects ofsediment routing on the downstream distribution of sedimentflux The theoretical analysis required in addition to prob-ability distributions of erosion (eg Figure 7) probabilitydistributions of transport capacities transport velocities andparticle attrition (Benda and Dunne 1997b) Sediment fluxfrom individual hillslopes and small low-order basins is pre-

dicted to be punctaform and thereforestrongly right-skewed similar to theOCR example (Figure 7) The channelobtains sediment over long periodsfrom such numerous right-skewed dis-tributions in small headwater basinsThe accumulation of sediment fromthe multitude of sediment sources overtime causes the flux distribution toevolve into more symmetrical forms ademonstration of the central limittheorem (Figure 8 also see the moregeneral form in Figure 4) The modelalso predicts that the magnitude ofsediment perturbations decreasesdownstream in conjunction with a cor-responding increase in their numberIncrease in perturbation frequency isdue to an increasing number and there-fore probability of sediment releasesdownstream A decrease in magni-tude of the perturbation (ie differ-ence between minimum and maximumvalues) is due to several factors in-cluding 1) particle attrition (break-down) that increasingly damps sedi-ment signals with distance traveled2) a larger and persistent supply andtherefore store of gravel 3) diffusion ofdiscreet sediment pulses due to selec-tive transport and temporary storagein bars and 4) increasing channelwidth (Benda and Dunne 1997b)

Episodic introduction of sediment cancreates pulses or waves of sedimentthat migrate thought the network caus-ing changes in channel morphologyincluding variations in channel bedelevation sinuosity substrate sizespools etc (Gilbert 1917 Madej andOzake 1996 Benda 1990 Miller andBenda in press) In addition fluctua-tions in sediment supply may also

cause changes in sediment storage at a particular locationsuch as near fans or other low-gradient areas and these havebeen referred to as stationary waves (Benda and Dunne 1997b)Migrating or stationary waves create coarse-textured cut andfill terraces (Nakamura 1986 Roberts and Church 1986Miller and Benda in press) Hence the predicted fluctuationsin sediment supply by the model (Figure 8) have morphologicalconsequences to channels and valley floors and therefore on tothe biological communities that inhabit those environments

Example Four Large Woody Debris

The supply and storage of large woody debris (LWD) in streamsis governed by several landscape processes that occur punctu-ated in time over decades to centuries including by fireswindstorms landslides stand mortality and floods A woodbudgeting framework (Benda and Sias 1998 in press) wasused to evaluate how those landscape processes constrainlong-term patterns of wood abundance in streams

In the example below the effect of two end member fire

Rel

ativ

e F

requ

ency

(co

lum

ns)

Cum

ulative frequency (lines)

00

01

02

03

04

15 - 20 30 - 35 45 - 50 60 - 6500

04

06

08

150 500

150 500

Fire recurrence interval (years)

B

LWD storage (vu per 100 m reach)

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Year (first fire at year 500)

LWD

sto

rage

(vu

1

00 m

)A FIRES

(500-yr cycle)

Stand mortality

FIRES(150-yr cycle)

150500

Fire recurrence interval (years)

05 10

02

0 - 5 75 - 80

Figure 9 (A) Patterns of wood storage for fire cycles of 150 and 500 years Gradualincreases in storage represent chronic stand mortality The magnitude of the abruptpulses of wood storage is governed by the amount of standing biomass (controlled by forestage) and the time interval of the toppling of fire-killed trees (40 years in this example) (B)More frequent fires result in a compressed range of variability and a shift in thedistribution towards lower wood volumes (solid bars represent the 500-year cycle) Lessfrequent fires shift the distribution of LWD volumes towards the right into higher volumesThese patterns indicate the potential for significant differences in LWD storage alongclimatic gradients in the Pacific Northwest region (east to west and north to south)

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

12 WMC Networker Summer 2001

regimes on variability in wood loading was examined a 500-year cycle associated with the wettest forests and a 150-yearcycle associated with more mesic areas in the southern andeastern parts of the Pacific Northwest region To reducecomplexity the analysis of fire kill and subsequent forestgrowth applied a series of simplifications (refer to Benda andSias 1998 in press)

The magnitude of wood recruitment associated with chronicstand mortality is significantly higher in the 500-year cyclebecause the constant rate of stand mortality is applied againstthe larger standing biomass of older forests (Figure 9) Fur-thermore fire pulses of wood are significantly greater in the500-year fire cycle because of the greater standing biomassassociated with longer growth cycles Hence longer fire cyclesyield longer periods of higher recruitment rates and higherpeak recruitment rates post fire Post-fire toppling of trees inthe 500-year cycle however accounts for only 15 of the totalwood budget (in the absence of other wood recruitment pro-cesses) In contrast drier forests with more frequent fires (eg150 year cycle) have much longer periods of lower wood recruit-ment and lower maximum post-fire wood pulses (Figure 8)Because the average time between fires in the 150-year cycleis similar to the time when significant conifer mortality occursin the simulation (100 yrs) the proportion of the total woodsupply from post-fire toppling of trees is approximately 50Hence stand-replacing fires in drier forests play a much largerrole in wood recruitment compared to fires in wetter forestsAlthough the range of variability of wood recruitment in therainforest case is larger the likelihood of encountering moresignificant contrasts (ie zero to a relatively high volume) is

higher in drier forests (Figure 8) Refer to Benda and Sias(1998 in press) for LWD predictions involving bank erosionlandsliding and fluvial transport

Potential For Developing New General LandscapeTheories

There is currently a paucity of theory at the spatial andtemporal scales relevant to ecosystems in a range of water-shed scientific disciplines (ie predictive quantitative under-standing) including hydrology (Dooge 1986) stream ecology(Fisher 1997) and geomorphology (Benda et al 1998) Henceour understanding of disturbance or landscape dynamics hasremained stalled at the ldquoconcept levelrdquo as evidenced by theproliferation of watershed-scale qualitative concepts acrossthe ecological and geomorphological literature including RiverContinuum concept (Vannote et al 1980) Patch Dynamicsconcept (Townsend 1989) Ecotone concept (Naiman et al1988) Flood Pulse concept (Junk et al 1988) Natural FlowRegime concept (Poff et al 1989) Habitat Template concept(Townsend and Hildrew 1994) Process Domain concept (Mont-gomery 1999) and hierarchical classification of space ndash timedomains (Frissel et al 1986 Hilborn and Stearns 1992)Although these concepts represent breakthroughs and innova-tions they have also promoted the arm waving aspect of scalein the watershed sciences whether looking over a ridge in thefield or in scientific or policy meetings It is also the primaryreason why natural resource management and regulatory com-munities prefer more reductionist and small scale approachesin their work

Any scientific field can encompass a range of different predic-tive theories that apply to differentcombinations of space and timescales Taking geomorphology as anexample (see Benda 1999 Figure 1)starting at the largest scales theo-ries of basin and channel evolution(eg Smith and Bretherton 1972Willgoose et al 1991) are applicableto watersheds or landscapes over 106to 108 years Because of the largetime scales involved they have lim-ited utility for natural resource man-agement and regulation At smallerscales there are theories dealing withlandsliding (Terzaghi 1950) and sedi-ment transport (duBoys 1879 Parkeret al 1982) At this level the spatialscale is typically the site or reachand changes over time intervals ofdecades to centuries are typically notconsidered Theories at this scale areuseful to studies of aquatic ecology atthe reach or grain scale (bed scouretc) and again are of limited utilityfor understanding managing or regu-

Figure 10 New landscape ndash scale theories will use distribution parameters of thingssuch as climate topography and vegetation with when integrated will yieldprobability distributions of output variables such as erosion and sediment supply (orLWD see Figure 9)

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 13

lating watershed behavior at larger scales Between these twosets of scales is a conspicuous absence of theoretical under-standing that addresses the behavior of populations of land-scape attributes over periods of decades to centuries Thematerial presented in this article is designed to eventually fillthat theoretical gap

At the mid level in such a theory hierarchy knowledge will takethe form of relationships among landscape processes (definedas distributions) and distributions of mass fluxes and channeland valley environmental states (Benda et al 1998) The newclass of general landscape theories will be based on processinteractions even if at coarse grain resolution among cli-matic topographic and vegetative distributions (Figure 10)The resultant shapes of probability distributions of fluxes of

sediment water or LWD will have consequences for riparianand aquatic ecology (Benda et al 1998) and therefore resourcemanagement and regulation

Increasing Scale Implications for WatershedManagementThe topic of increasing scale in resource management is virtu-ally unexplored and it will only be briefly touched on hereRecently forest managers have been promoting the establish-ment of older forest buffers along rivers and streams to main-tain LWD and shade and for maintaining mature vegetationon landslide prone areas for rooting strength Often themanagement andor regulatory target is mature or old growthforests at these sites However in a naturally dynamic land-scape forest ages would have varied in space and time The

Stand Age (yrs)0 25 50 75 100 150 200

Zero Order

0

2

4

6

8

10

175125

o

f Ch

ann

el L

eng

th

First Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Second Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Third Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Fourth Order

0

2

4

6

8

10

Stand Age (yrs)0 25 50 75 100 150 200175125

o

f Ch

ann

el L

eng

th

Figure 11 Probability distributions of riparian forest ages including for landslide sites The average proportion of channel lengthwith forest stands of a particular age using 25-year bins up to 200 years (forests greater than 200 years not shown) was tabulatedfor zero-order (ie landslide sites) through fourth order These predicted histograms indicate that on average the proportion of thechannel length containing trees less than 100 years old varies from 30 (zero order) 24 (first-order) to 15 (fourth order) Thedecreasing amount of young trees with increasing stream size is a consequence of the field estimated susceptibility of fires Firefrequency was the highest on ridges and low-order channels (~175 yrs) and the lowest (~400 yrs) on lower gradient and wide valleyfloors

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

14 WMC Networker Summer 2001

temporal variability in forest coveris illustrated in Figure 6 Howeverfire regimes are sensitive to topo-graphic position (Benda et al 1998Figure 115) and hence the propor-tion of stream channels and land-slide sites bordered by mature orold forests would vary with topog-raphy

The variability in forest ages in a200 km2 watershed located insouthwest Washington was inves-tigated using a fire model Althoughthe long-term probability distribu-tion of forest ages predicted by themodel was positively skewed inaccordance with existing theory(Johnson and Van Wagner 1984)and as illustrated in Figure 6(B)distributions of forest ages variedwith topographic position or streamsize The distributions for forestages up to year 200 for a range ofchannel sizes in western Washing-ton indicated that there was ahigher likelihood of younger ageclasses in landslide prone sites (iezero-order) and in the smalleststeepest streams (Figure 11) Anincreasing amount of the probabil-ity distribution of age classesshifted right into older forests withincreasing stream size This is dueto a lower fire frequency in lowergradient and wider valley floorscompared to steeper hillslopes and channels The simulationsindicated that natural forests cannot be represented ad-equately by a specific age class but rather by a distribution ofvalues the shape of which varies with region and topographicposition

This type of information could conceivably be used to craftfuture management strategies that mimicked the age distri-bution of forests along certain topographic positions Forexample the temporal distribution of forest ages at landslidesites shown in Figure 11 could according to the ergodic hypoth-esis illustrated in Figure 5 be considered to reflect the spatialdistribution of vegetation age classes in any year across alandscape Figure 11 suggests that approximately 18 oflandslide sites in the study area of southwest Washington wascovered with young less than 50 year old vegetation at anypoint in time Management strategies focused on landslidereduction might use such information to manage differentareas with different timber harvest rotations

Increasing Scale Implications for RegulationBecause landscape behavior is complex in space and timeevaluating human impacts to terrestrial and aquatic habitatscan pose a difficult problem The use of distributions outlinedin this article may provide fertile ground for developing newrisk assessment strategies First and most simply the sever-ity of environmental impacts could be evaluated according tothe degree of observed or predicted shifts in frequency or

probability distributions in space or time under various landuses For example process rates or environmental conditionsthat fall onto tails of distributions could be considered exceed-ing the range of variability (eg a sediment depth of 15 m inchannels of 25 km2 or greater in Figure 8) Second sincedistributions define probability of event occurrence probabil-ity density functions of certain landscape processes couldunderpin risk assessment For example the chance of encoun-tering 20 landslides in any year at a scale of 25 km2 is less than1 and such an erosional condition probably requires eitherwidespread fire or wholesale clearcutting (Figure 7) The thirdand the most challenging avenue would involve estimatingthe level of disturbance (ie its regime) that is optimal forcertain types of ecosystems and species These potentialtechniques would presumably be based on a body of theoryapplicable at large scales

To follow up on the first potential option some environmentalsystems are so complex that simply knowing whether systemslie inside or outside their range of natural variability mightprove useful particularly in the absence of more detailedunderstanding This approach is commensurable with therange-of-variability concept (Landres et al 1999) and hasbeen applied recently in the global climate change debate(EOS 1999) Because of human civilization almost no naturalsystems lie within their pre-human natural range of variabil-ity However the concept might prove useful in decipheringhow far out of whack an environmental system is or to comparecertain types of impact with others (ie logging vs urbaniza-

Figure 12 Distributions may offer new avenues for environmental problem solving Eitherthe degree of shifts in distributions of watershed attributes in space or time might provideinsights into environmental problems (A) Even young vegetation occurred naturally insmall areas temporally persistent timber harvest in upper watersheds can lead to a systembeing outside its range of natural variability in time (B) Over short time periods howeverspatially pervasive timber harvest may also cause a system to shift outside its natural rangeof variability in space (C) Other forms of land uses such as dams or diking will causesignificant and persistent shifts in probability distributions of flooding sedimentation andfloodplain construction The degree of shifts among the different watershed attributes mayoffer a means to consider the cumulative effects over space and time

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 15

ReferencesAllen TFH and TB Starr 1982 Hierarchy Perspectives forEcological Complexity University of Chicago Press Chicago310ppBenda L 1994 Stochastic geomorphology in a humidmountain landscape PhD dissertation Dept of GeologicalSciences University of Washington Seattle 356pBenda L and T Dunne 1997a Stochastic forcing of sedimentsupply to the channel network from landsliding and debrisflow Water Resources Research Vol 33 No12 pp2849-2863Benda L and T Dunne 1997b Stochastic forcing of sedimentrouting and storage in channel networks Water ResourcesResearch Vol 33 No12 pp 2865-2880Benda L and J Sias 1998 Wood Budgeting LandscapeControls on Wood Abundance in Streams Earth SystemsInstitute 70 ppin press CJFASBenda L D Miller T Dunne J Agee and G Reeves 1998Dynamic Landscape Systems Chapter 12 in River Ecologyand Management Lessons from the Pacific CoastalEcoregion Edited byNaiman R and Bilby R Springer-Verlag Pp 261-288Benda L (1999) Science VS Reality The Scale Crisis in theWatershed Sciences Proceedings Wildland HydrologyAmerican Water Resources Association Bozeman MT p3-20Botkin D B 1990 Discordant Harmonies A New Ecology forthe Twenty-First Century Oxford University Press NewYork 241ppBrunsden D and J B Thornes (1979) ldquoLandscape sensitivityand changerdquo Transactions Institute of British GeographersNS 4 485-515Caine N 1990 Rainfall intensity ndash duration control onshallow landslides and debris flows Geografika Annaler62A23-27du boys P F D (1879) lsquoEtudes du regime et lrsquoaction exerceepar les eaux sur un lit a fond de graviers indefinementaffouliablersquo Annals des Ponts et Chaussees Series 5 18 141-95Dooge J C I 1986 Looking for hydrologic laws WaterResources Research 22(9) pp46S-58SDunne T 1998 Critical data requirements for prediction oferosion and sedimentation in mountain drainage basinsJournal of American Water Resources Association Vol 34pp795-808EOS 1999 Transactions American Geophysical Union V 80No 39 September (Position statement on climate change andgreenhouse gases)Fisher S G 1997 Creativity idea generation and thefunctional morphology of streams J N Am Benthl Soc 16(2)305-318Frissell C A W J Liss W J Warren and M D Hurley1986 A hierarchical framework for stream classificationviewing streams in a watershed context EnvironmentalManagement 10 pp199-214Gell-Man M 1994 The Quark and the Jaguar Adventures inthe Simple and Complex W H Freeman and Company NewYork 329ppGilbert GK 1917 Hydraulic mining debris in the SierraNevada US Geological Survey Prof Paper 105 154pHack JT and Goodlett JC 1960 Geomorphology and forestecology of a mountain region in the central AppalachiansProfessional Paper 347 US Geological Survey Reston Va66ppHeede BH 1988 Sediment delivery linkages in a chaparrel

tion vs dams)

The distribution parameter an index sensitive to changingspatial and temporal scales (eg Figures 1 ndash 4) allows envi-ronmental changes to be registered as shifts in distributionsAn example of how a forest system might be managed outsideits range of natural variability in time is given in Figure 12 Insmall watersheds (~40 km2) the frequency distribution offorest ages observed at any time is naturally variable becauseof episodic stand-replacing fires (Figure 6) Although fire isnot equivalent to timber harvest relatively young forestsfollowing timber harvest would have had an approximatecorollary in post fire forests with respect to things such assubsequent LWD recruitment and soil rooting strength In anatural forest however the distribution of forest ages wouldgradually shift towards the older ages over time If timberharvest were to persist over several rotations the distribu-tions would persist in the youngest age classes and thereforereveal management beyond the natural range of variability intime in that limited context

Figure 12 also illustrates how the natural range of variabilitycan be exceeded in space At large spatial scales (landscape)the distribution of forest ages is predicted to be significantlymore stable over time (Figure 6) If timber harvest occurs overthe entire 20000 km2 OCR over a relatively short period oftime (a few years to a few decades) the left-shifted age distri-bution that would result represents a forest that lies outsideitrsquos range of natural variability in space in any year or shortperiod of time (Figure 10)

The degree of distribution shift might prove useful whenprioritizing where to focus limited resources in stemmingenvironmental impacts that originate from a range of landuses For example the degree of shift in the age distribution offorests in upland watersheds due to logging could be con-trasted with a distribution shift in flooding sedimentationand floodplain construction due to diking in the lowlands(Figure 12)

ConclusionsA focus on small scales and a desire for accuracy and predict-ability in some disciplines has encouraged scientific myopiawith the landscape ldquobig picturerdquo often relegated to arm wavingand speculation This could lead to several problems some ofwhich are already happening First there may be a tendencyto over rely on science as the ultimate arbitrator in environ-mental debates at the wrong scales This can lead to thecreation of unrealistic management and regulatory policiesand the favoring of certain forms of knowledge over otherforms such as promoting quantitative and precise answers atsmall scales over qualitative and imprecise ones at largescales Second science may be used as a smoke screen tojustify continuing studies or management actions in the faceof little potential for increased understanding using existingtheories and technologies at certain scales Third science canbe used as an unrealistic litmus test requiring detailed andprecise answers to complex questions at small scales when nosuch answers at that scale will be forthcoming in the nearfuture Finally the persistent ideological and divisive envi-ronmental debates that have arisen in part based on theproblems of scale outlined above may undermine the publicrsquosbelief in science It may also weaken the support of policymakers for applying science to environmental problems Toattack the problem of scale in watershed science manage-ment and regulation will require developing new conceptualframeworks and theories that explicitly address interdiscipli-nary processes scientific uncertainty and new forms of knowl-edge-

Earth Systems Institute is a private non profit think tanklocated in Seattle WA (wwwearthsystemsnet)

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

16 WMC Networker Summer 2001

watershed following wildfire Environmental ManagementVol 12(3)349-358Hogan DL SA Bird and MA Haussan 1995 Spatial andtemporal evolution of small coastal gravel-bed streams theinfluence of forest management on channel morphology andfish habitats 4th International Workshop on Gravel-bedRivers Gold Bar WA August 20-26Johnson E A and Van Wagner C E 1984 The theory and useof two fire models Canadian Journal Forest Research 15214-220Junk W Bayley P B and Sparks R E 1989 The flood-pulseconcept in river-floodplain systems Canadian SpecialPublication of Fisheries and Aquatic Sciences 106 110-127Kellert S H 1994 In the Wake of Chaos UnpredictableOrder in Dynamical Systems University of Chicago PressChicago 176ppLackey R 1998 Ecosystem Management DesperatelySeeking a Paradigm in Journal of Soil and Water Conserva-tion 53(2) pp92-94Landres P B Morgan P and Swanson F J 1999 Overviewof the use of natural variability concepts in managing ecologi-cal systems Ecological Applications 9(4) 1179 ndash 1188Long C J Fire history of the Central Oregon Coast RangeOregon 9000 year record from Little Lake MS thesis 147 ppUniv of Oreg EugeneMaurer B A (1999) Untangling Ecological ComplexityUniversity of Chicago Press ChicagoMcIntrye L 1998 Complexity A Philosopherrsquos ReflectionsComplexity 3(6) pp26-32Meyer G A Wells S G and Timothy Jull A J 1995 Fireand alluvial chronology in Yellowstone National Parkclimatic and intrinsic controls on Holocene geomorphicprocesses GSA Bulletin V107 No10p1211-1230Miller D J and Benda L E in press Mass Wasting Effectson Channel Morphology Gate Creek Oregon Bulletin ofGeological Society of AmericaMongomery D R 1999 Process domains and the rivercontinuum Journal of the American Water Resources Associa-tion Vol 35 No 2 397 ndash 410Nakamura R 1986 Chronological study on the torrentialchannel bed by the age distribution of deposits ResearchBulletin of the College Experiment Forests Faculty ofAgriculture Hokkaido University 43 1-26Naiman R J Decamps H Pastor J and Johnston C A1988 The potential importance of boundaries to fluvialecosystems Journal of the North America BenthologicalSociety 7289-306Naiman R J and Bilby R E 1998 River Ecology andMangement Lessons from the Pacific Coastal EcoregionSpringer-Verlag 705ppParker G Klingeman PC and McLean 1982 Bedload andsize distribution in paved gravel-bed streams J HydraulEng 108 544-571Pickett STA and PS White eds 1985 The ecology ofnatural disturbance and patch dynamics Academic Press NewYork New York USAPickup G Higgins R J and Grant I 1983 Modellingsediment transport as a moving wave - the transfer anddeposition of mining waste Journal of Hydrology 60281-301Poff N L Allen J D Bain M B Karr J Prestegaard KL Richter BD Sparks RE and Stromberg JC 1997 Thenatural flow regime A paradigm for river conservation andrestoration Bioscience V47 No11 769-784 Robichaud P R and Brown R E 1999 What happened afterthe smoke cleared Onsite erosion rates after a wildfire in

Eastern Oregon Proceedings Wildland Hydrology AmericanWater Resources Association Ed D S Olsen and J PPotyondy 419-426Reeves G Benda L Burnett K Bisson P and Sedell J1996 A disturbance-based ecosystem approach to maintainingand restoring freshwater habitats of evolutionarily significantunits of anadromous salmonids in the Pacific NorthwestEvolution and the Aquatic System Defining Unique Units inPopulation Conservation Am Fish Soc Symp 17Reid L 1998 Cumulative watershed effects and watershedanalysis Pacific Coastal Ecoregion Edited by Naiman R andBilby R Springer-Verlag pp476-501Rice RM 1973 The hydrology of chaparral watersheds ProcSymp Living with chaparral Sierra Club Riverside Califpp27-33Rodriguez-Iturbe I and J B Valdes 1979 The geomorphicstructure of hydrologic response Water Resources Research15(6) pp1409 ndash 1420Roberts R G and M Church 1986 The sediment budget inseverely disturbed watersheds Queen Charlotte RangesBritish Columbia Canadian Journal of Forest Research161092-1106Roth G F Siccardi and R Rosso 1989 Hydrodynamicdescription of the erosional development of drainage pat-terns Water Resources Research 25 pp319-332Smith TR and Bretherton F P 1972 Stability and theconservation of mass in drainage basin evolution WaterResources Research 8(6) pp1506-1529Swanson FJ R L Fredrickson and F M McCorison 1982Material transfer in a western Oregon forested watershed InRL Edmonds (ed) Analysis of coniferous forest ecosystems inthe westernUnited States Hutchinson Ross Publishing Co StroudsburgPenn pp233-266Swanson FJ Kratz TK Caine N and Woodmansee R G1988 Landform effects on ecosystem patterns and processesBioscience 38 pp92-98Swanson F J Lienkaemper G W et al 1976 Historyphysical effects and management implications of largeorganic debris in western Oregon streams USDA ForestServiceTeensma PDA 1987 Fire history and fire regimes of thecentral western Cascades of Oregon PhD DissertationUniversity of Oregon Eugene Or 188pTerzagi K 1950 Mechanism of landslides In Paige S edApplication of geology to engineering practice Berkeley VolGeological Society of America 82-123Trimble S W 1983 Changes in sediment storage in the CoonCreek basin Driftless Area Wisconsin 1853 to 1975 Science214181-183Vannote R L G W Minshall K W Cummins J R Sedelland C E Cushing 1980 The river continuum conceptCanadian Journal of Fisheries and Aquatic Sciences 37pp130-137Waldrop M M 1992 Complexity The Emerging Science atthe Edge of Order and Chaos Touchstone Books Simon andSchuster New York 380ppWeinberg G M 1975 An Introduction to General SystemsThinking Wiley-Interscience New YorkWohl EE and Pearthree PP 1991 Debris flows as geomor-phic agents in the Huachuca Mountains of southeasternArizona Geomorphology 4 pp273-292Willgoose G Bras RL and Rodriguez-Iturbe I 1991 Acoupled channel network growth and hillslope evolutionmodel 1 Theory Water Resources Research Vol 2 No1pp1671-1684

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 17

Timber Harvest and Sediment Loads in Nine NorthernCalifornia Watersheds Based on Recent Total Maximum Daily Load(TMDL) StudiesContinued from Page 1

source of sediment delivered to stream channels (Swansonet al 1982 Dietrich et al 1986 Dietrich et al 1998 Roeringet al 1999) It is generally hypothesized that long-termsediment production rates in this region are geologicallycontrolled by rates of tectonic uplift which influencestopography and certain mechanical properties of thebedrock and therefore its sensitivity to land use (Ahnert1970 Summerfield and Hulton 1994 Leeder 1991)

Shallow landslides occur in shallow soils and are typicallydriven by transient elevated pore pressures in the soilsubsurface resulting from intense precipitation during astorm event (Campbell 1975 Reneau and Dietrich 1987)Since pore pressures are drainage area-dependent conver-gent areas on hillslopes are more likely to experience soilsaturation conditions and reduced shear strength and thushave higher probability of generating a shallow landslideunder natural conditions (Wilson and Dietrich 1987Dietrich et al 1992 1993 Montgomery and Dietrich 1994Tucker and Bras 1998) Substantial natural landslidinghas been triggered during several large storms during therecent past (eg 1964 1973 1975 1986 1992 and 1997)However it should be noted that any changes to hillslopehydrology such as increased inputs of rainfall and runoff inthe shallow subsurface due to forest management practicesgenerally results in increased frequency of occurrence ofshallow landsliding which leads to accelerated sediment

loading to stream channels alteration of channel morphol-ogy and degradation of stream habitat (Sidle et al 1985Sidle 1992 Dietrich et al 1993 Montgomery et al19982000)

Because few North Coast watersheds within the range ofthe coho salmon have been entirely unimpacted by pastlogging activities there have been very few publicationsaccurately associating specific forest practices with sedi-ment delivery This stems from the difficulty in findingundisturbed reference sites the long recurrence interval fornaturally induced large scale mass wasting events and thedecades-long residence time of in-channel sediment storagerequired for delivered stream sediment to move through thesystem (Dietrich and Dunne 1978 Madej 1995)

Harvest-related ldquoin-unitrdquo erosion generally takes thefollowing forms 1) accelerated shallow landsliding due toloss of root strength increase in ground moisture and lowerattentuation of rainfall inputs (Wilson and Dietrich 1987Dietrich et al 1992 Montgomery and Dietrich 1994Keppeler and Brown 1998 Montgomery et al 2000) 2)destabilized deep seated landsliding due to increasedground moisture and loss of landslide toe support (Swansonet al 1987 Miller 1995) and 3) increased overland flow(surface) erosion due to ground disturbance by forestmanagement In the Pacific Northwest the majority of

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

583581975-1990South Fork Trinity River

5427191981-1996South Fork Eel River

1277121952-1997Garcia River1

7210181955-1999Van Duzen River1

4044171954-1980Redwood Creek

613451975-1998Navarro River

3641241979-1999Ten Mile River

563951979-1999Noyo River

1940411976-1989Grouse Creek Watershed of SF Trinity

Natural4Other Forest Management

Related3

Harvest-Related2

Years of Study

TMDL Sediment Source Analysis

Table 1 Summary of TMDL sediment source analyses for nine watersheds within the rangeof the coho salmon (Oncorhynchus kisutch) in northern California Data reported reflectfraction of total sediment loading by land use association for periods indicated

Notes1 Timber harvest and road building contributions are assumed to have changed beforeand after the ZrsquoBerg Nejedley Forest Practice Act of 19732 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion3 Sources include road- and skid-related surface erosion and mass wasting4 Sources include mass wasting surface erosion and creep

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

18 WMC Networker Summer 2001

sediment budgets in managed watersheds are dominated byroad-related sediment inputs While road-related erosion isnot treated as a timber harvest-related source in thisreview in practice it should be treated as ldquoharvest-relatedrdquosince the majority of these roads support timber-harvestingactivities and were built for the purposes of timber hauling

Methods and Uncertainty In general sediment source analyses used in TMDLsprogress from estimates of actual or potential loading fromhillslopes and banks to receiving waters estimates of in-stream storage and transport of sediment to estimates ofthe net sediment discharge (or yield) from drainage basins(eg Reid and Dunne 1996 USEPA 1999a) The techniquesgenerally use aerial photo analysis of mass wasting andsurface erosion features GISndashDTM (Geographic InformationSystemndashDigital Terrain Model) analyses and limitedground surveys of geology and landslide characteristics andin-stream measures of sediment storage and transportAlthough geology and topography are variable across small(100 km2) watersheds within the range of the coho salmonstratification of the landscape sediment supply by geomor-phic terrains could be expected to yield similar rates ofproduction due to similarities in geology and topographyThese geomorphic terrains are generally defined by geologytectonics topography geomorphology soils vegetation andprecipitation Further stratification within geomorphicterrains by land use can be reasonably assumed to revealdifferences in sediment production by comparing areas withand without particular human activities

Although the degree of uncertainty greatly depends upon themethodology used the range of uncertainty in sediment

source analyses is generally on the order of 40ndash50 (Rainesand Kelsey 1991 Stillwater Sciences 1999) Methodologicalconstraints (eg estimates of landslide frequency arealextent depth age bulk density estimates of landslidedelivery ratio and natural temporal variability in erosion-triggering storm events) suggest that this uncertainty maybe too high to reliably detect differences between land usesor recent changes in land use practice such as those intro-duced in 1973 under the ZrsquoBerg Nejedley Forest Practice Act(FPA) of 1973 (CCR 14 Chapters 4 and 45)

ApproachDespite the limitations described above the approvedmethodology for TMDLs (USEPA 1999a) has been used todevelop sediment source analyses for several watershedswithin the range of the coho salmon in northern Californiaand can also be used to assess the relative impacts oftimber harvesting activities on sediment loading in streamchannels In order to make rational comparisons betweenthe published TMDLs three factors need to be addressedFirst sediment yield data in many publications encompasslong periods during which forest practices changed Wherepossible we selected sediment source data that wereprovided for the post-1973 FPA period to more accuratelyreflect current forest practices Second land use aggrega-tions differ among published TMDLs For example road-related mass wasting is aggregated within timber harvest-related sediment production in some studies and is aseparate sediment source in others Third the difference inperiodicity of triggering storms during the time frames usedto assess sediment sources in individual TMDLs was notaddressed Note that this analysis uses only existing

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

584041Maximum

473518Mean

Post-Forest Practice Act Sediment Sources from Finalized TMDLs 4 (n=4)

Sediment Sources from All TMDL Data Presented in Table 1 (n=9)

1139Standard Error

764Standard Error

1245Minimum

19275Minimum

727741Maximum

453817Mean

Natural3Other Forest Management

Related2Harvest-Related1

Table 2 Human and natural sediment contributions to total sediment loading within the range of the cohosalmon (Oncorhynchus kisutch) in northern California for all TMDLs and those TMDLs with informationavailable after the 1973 Forest Practice Act

Notes 1 Sources include hillslope mass wasting and ldquoin-unitrdquo surface erosion 2 Sources include road- andskid-related surface erosion and mass wasting 3 Sources include mass wasting surface erosion and creep4 Of the nine TMDLs that have been finalized only four provide post-Forest Practice Act sediment sourceassessments Data include Grouse Creek Watershed of SF Trinity Noyo River South Fork Eel River andSouth Fork Trinity River

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 19

results of sediment source analyses for the total period ofrecord Where possible we have separated results to showsediment source contributions since the passage of the 1973FPA Although a more thorough meta-analysis of thebackground data of currently published TMDLs is notpossible without considerable effort we provide both totalestimates of natural vs human-related sediment loading to

stream channels as well as a timber harvest-relatedestimate only

ResultsBased upon our initial review of the sediment sourceanalyses reported here two analyses indicate a reduction intotal sediment loading after the FPA of 1973 For the

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

1955-1999

1979-1999

1975-1990

1981-1996

1954-1997

1979-1999

1975-1998

1976-1989

1952-1997

Period of Analysis

1115

311

2414

1784

738

293

816

147

296

Area (km2)

4282194Eastern Belt Franciscan Complex metamorphic and metasedimentary rocks

South Fork Trinity River

46326704Coastal and Central Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

South Fork Eel River

88469533Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Garcia River(1)

28196(4)700(4)Central and Eastern Belt Franciscan Complex Tertiary marine sedimentary rocks

Van Duzen River(1)

6011051834Central Belt Franciscan Complex metasedimentary rocks

Redwood Creek(1)

39290745Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Navarro River

64195303Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Ten Mile River

44114258Coastal Belt Franciscan Complex Tertiary to Cretaceous marine sedimentary rocks

Noyo River

81201249(3)Eastern Belt Franciscan Complex pre-Cretaceous metasedimentary rocks

Grouse Creek Watershed of SF Trinity

Ratio of Anthropogenic

to Total Sediment

Mean Annual Sediment Yield Associated with

Timber Harvest and Roads(2)(tonskm2-yr)

Total Mean Annual

Sediment Yield

(tonskm2-yr)

Geologic Unit TypesTMDL

Sediment Source Analysis

Table 3 Sediment yields for all terrain types in nine northern California watersheds

Notes 1) Timber harvest and road building contributions are assumed to have changed before and after the ZrsquoBerg NejedleyForest Practice Act of 1973 2) Sources include hillslope mass wasting skid trail surface erosion and road-related surface erosion3) Excludes stream bank erosion due to data uncertainty 4) Estimated by assuming bulk density of 18 tonsm3

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

20 WMC Networker Summer 2001

period of 1954ndash1980 the Redwood National and StateParks (1997) estimated a total sediment load in theRedwood Creek basin of 1883 tonskm2-yr whereas thisrate dropped to 1070 tonskm2-yr for the period 1981ndash1997(USEPA 1998c) In the South Fork of the Trinity RiverRaines (1998) estimated that the 1944ndash1975 sedimentproduction rate (432 tonskm2-yr) fell to 194 tonskm2-yr forthe period 1976ndash1990 These correspond to a 57 and 45decline in total sediment production rates respectivelyindicating that the pre- and post-FPA production aresubstantially different However a simple comparison suchas this does not account for potential differences in thefrequency and magnitude of large storms that serve astriggering events for sediment delivery Regardless it is onthis basis that we attempted to include as much post-FPAdata as possible to better assess the current effects oftimber harvesting within the range of the coho salmonTable 1 presents the proportion of the total sediment loadsin nine North Coast basins that may be attributed totimber harvesting other human-causes such as roadbuilding and natural sources Table 2 provides the range ofdata associated with human-related activities and natural

sources for the period before and after the 1973 FPA Table3 provides information on sediment yields for each basinThe results are discussed for each basin below

Garcia River The Garcia River TMDL includes the years1952ndash1997 (USEPA 1998a) For the entire period of recordPacific Watershed Associates (1997) estimated an averagesediment production rate of 533 tonskm2-yr within theGarcia River basin (Table 3) Road building activitiesdominate the sediment budget for the period of record(Figure 1) Grouping the source analysis data by natural andhuman caused processes results in an anthropogenicassociation of 89 of all sediment production within thebasin Approximately 12 of this total is directly attributedto timber harvest-related activities (Table 1)

Grouse Creek sub-Watershed of SF Trinity Thesediment budget for the Grouse Creek sub-watershed withinthe South Fork Trinity River (Raines 1998) was developedfor the Six Rivers National Forest It covers a period from1960ndash1989 and represents the only portion of the SouthFork Basin where a detailed sediment budget wascompleted (USEPA 1998b) For the entire period of record

Raines (1991) estimated an average sediment productionNATURAL Mass wasting

12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

NATURAL Mass wasting12

HARVEST Mass Wasting12

OTHER MGMT Mass WastingRoads

35OTHER MGMT Road

Surface Erosion 3

OTHER MGMT Road and SkidTrail Crossings and Gullies

from Diversions on Roads andSkid Trails

38

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

OTHER MGMT Masswasting Roads

292

OTHER MGMT RoadSurface Erosion including

X-ing washouts104

HARVEST Mass Wasting204

HARVEST In-Unit surface erosion

208

NATURAL Mass wasting190

NATURAL Surface erosion(fire grazing)

01

NATURAL Stream bankerosion 00

NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2 NATURAL shallowlandslides

9

NATURAL deep seatedlandslides

5

NATURAL gullies13

NATURAL bank erosion3

NATURAL innergorgestreamside

delivery31

OTHER MGMT Road-Stream crossing failures

7

OTHER MGMT Road-related mass wasting

6

OTHER MGMT Road-related gullying

6

HARVEST Mass Wasting3

OTHER MGMT Road-related surface erosion

13

OTHER MGMT Vineyarderosion

2

HARVEST In-Unitldquosurface erosion

2

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

26

OTHER MGMT Masswasting Railroads

1 NATURAL Mass wasting

15

NATURAL Surface erosion11

NATURAL Stream bankerosion31

OTHER MGMT Road-related mass wasting

11

HARVEST Mass Wasting3

Figure 1 Sediment Contributions to the Garcia River (1952-1997)Source Pacific Watershed Associates 1997

Figure 2 Sediment Contributions to the Grouse Creek Sub-Watershed of the South Fork Trinity River (1976-1989)Source Raines 1998

Figure 3 Sediment Contributions to the Navarro River(1975-1998) Source Navarro River TMDL (USEPA 2000a)

Figure 4 Sediment Contributions to the Noyo River (1979-1999) Source Mathews and Associates 1999

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 21

rate of 1750 tonskm2-yr within the Grouse Creekwatershed Grouping the source analysis data by naturaland human caused processes results in an anthropogenicassociation of 50 of all sediment production within thebasin (Raines 1991) For the years 1976ndash1989 (post-FPA)Raines (1998) assumed negligible natural stream bankerosion resulting in approximately 81 of the total averagesediment yield (249 tonkm2-yr) being related to all humancauses combined (Table 3) Approximately 41 of this totalmay be directly attributed to timber harvest-relatedactivities in this relatively small (147 km2) sub-basin(Figure 2 Table 1)

Navarro River Although the public review draft of theNavarro River TMDL (USEPA 2000a) does not providecitations for its sediment source analysis the sedimentsource analysis provided focused upon rates of sedimentyield that have occurred in the recent past (ie past twentyyears) For the estimated time period of this analysis(1975ndash1998) 61 of the sediment loading is attributed tonatural sources (Figure 3) Total sediment load was 745tonskm2-yr (Table 3) of which 34 may be attributed toother forest management activities (Figure 3) and only 5may be directly attributed to timber harvest-relatedactivities in this 816 km2 basin (Table 1)

Noyo River In the extensive sediment source analysis forthe Noyo River TMDL (USEPA 1999b) Matthews ampAssociates (1999) evaluated landsliding throughout thewatershed using 124000 scale aerial photographs for theyears 1942 1952 1957 1963 1965 1978 1988 1996 and1999 The 1942 aerial photos were assumed to give asnapshot of landscape events occurring over a 10-year period(ie back to 1933) The sediment yields for 1933ndash1957 inthe Noyo River watershed (85 tonskm2-yr) captures a periodof low intensity logging between old growth harvest (ca1900) and second growth harvest (post 1950s) For therecent period of this analysis (1979ndash1999) approximately44 of the total sediment loading may be attributed to allhuman-related activities combined (Figure 4) Total

sediment load in the recent past was 258 tonskm2-yr (Table3) of which only 5 may be directly attributed to timberharvest-related activities in this 293 km2 basin (Table 1)

Redwood Creek Although several sediment sourceanalyses were used in the Redwood Creek TMDL (USEPA1998c) detailed information required to separate sedimentsources by process and causality was unavailable for thepre- and post-FPA periods For the entire period of record(1954ndash1997) the Redwood Creek Watershed Analysisestimates a load of 1834 tonskm2-yr and also separatesthe sediment yields by process (Redwood National andState Parks 1997) (Table 3) Approximately 61 of thetotal sediment load during this period may be attributed to

all human-related activities combined (Figure 5) Seven-teen percent of the total sediment load may be attributed totimber harvest-related activities in this 738 km2 basin(Table 1)

South Fork Eel River The sediment source analysis(Stillwater Sciences 1999 USEPA 1999d) combined severalmethods to generate estimates of sediment in the SouthFork Eel Basin Earlier studies were summarized anddetailed photo analysis and fieldwork were conducted in theintensive study areas (ISAs) representative of various

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST Mass Wasting5

HARVEST In-Unitldquosurface erosion

9

OTHER MGMT Road-related mass wasting

7

OTHER MGMT Road-related surface erosion

32

OTHER MGMT Roads-washouts gullies

small slides23

NATURAL Mass wasting13

NATURAL Surface erosion11

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

HARVEST tributarylandslides

8

HARVEST Bare grounderosion

8

OTHER MGMT Roads ampSkid trails (incl

Silvicult agri public)15

OTHER MGMT Roadrelated tributary

landslides8

OTHER MGMT Road-related Gully erosion

22

NATURAL Stream Bankerosion12

NATURAL tributarylandslides

4

NATURAL Other Massmovements

(earthflowsblock slides)4

NATURAL Debris torrents2

NATURAL Mainstemlandslides

17

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

OTHER MGMT Road-related mass wasting

amp gullying22

HARVEST Shallowlandslides

17

NATURAL Soil Creep5

NATURAL Shallowlandslides

11

NATURAL Earthflow toesand associated gullies

38

HARVEST In-Unitldquosurface erosion

2

OTHER MGMT Road-related surface erosion

5

Figure 5 Sediment Contributions to the Redwood Creek(1954-1997) Source Redwood Creek TMDL (USEPA 1998cRedwood National and State Parks 1997)

Figure 6 Sediment Contributions to the South Fork EelRiver (1981-1996) Source Stillwater Sciences 1999

Figure 7 Sediment Contributions to the South Fork TrinityRiver (1975-1990) Source South Fork Trinity TMDL(USEPA 1998b Raines 1991)

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

22 WMC Networker Summer 2001

geomorphic terrains Existing data was supplemented byseveral modeling techniques and GIS data analysis wasused to extrapolate results from the ISAs into the entirebasin A major focus of the sediment source analysis was tobetter characterize the proportion of human-inducedsediment The USDA (1970) estimated a total sedimentproduction of 1951 tonskm2-yr with approximately 20 ofthe total from human-related activities from the period1942ndash1965 For the most recent time period (1981ndash1996)Stillwater Sciences (1999) estimated a basinwide rate ofsediment production of 704 tonskm2year with approxi-mately 46 of the total sediment load attributable to landuse activities (Table 3 Figure 6) Nineteen percent of thetotal sediment load may be attributed to timber harvest-related activities in this 1784 km2 basin (Table 1)

South Fork Trinity River The sediment source analysisfor the South Fork Trinity Basin TMDL (USEPA 1998b)provided detailed sediment budgets for all sub-basinsRaines (1998) estimated that for the 1944ndash1990 periodtotal sediment production was 407 tonskm2-yr with 9 ofthis total contributed by timber harvest related activitiesOver 86 of all sediment produced by landsliding was foundto be concentrated in areas of geologic instability andlogging and during major storms Sixty-three percent of allsediment production between 1944 and 1990 was generatedfrom 1961 to 1975 a period that included four major stormevents the completion of 74 of basin logging activity and80 percent of road building from 1960ndash1989 For the mostrecent time period (1976ndash1990) approximately 43 of the

total sediment load may be attributed to all human-relatedactivities combined (Figure 7) Total sediment productionfor this period was 194 tonskm2-yr of which 8 of the totalsediment load may be attributed to timber harvest-relatedactivities in this 2414 km2 basin (Tables 1 and 3)

Ten Mile River The public review draft of the Ten MileRiver TMDL (USEPA 2000b) compiled sediment sourcedata from a variety of sources including analyses conductedin adjacent basins (Mathews amp Associates 2000) For theperiod of this analysis (1979ndash1999) approximately 65 ofthe total sediment load may be attributed to all human-related activities combined (Figure 8) Total sediment

production was 303 tonskm2-yr of which 24 may beattributed to timber harvest-related activities in this 311km2 basin (Tables 1 and 3)

Van Duzen River The Van Duzen River TMDL (USEPA1999c) covers a period from 1955ndash1999 The sediment yieldrates were based on a study of the upper 60 of the water-shed conducted by Kelsey (1977) using aerial photos and astratified random sampling scheme for field validation

(Pacific Watershed Associates 1999) For the entire periodof record approximately 28 of the total sediment load maybe attributed to all human-related activities combined(Figure 9) Total sediment production was 700 tonskm2-yrof which 18 may be attributed to harvest-related activitiesin this 1115 km2 basin (Tables 1 and 3)

Discussion and ConclusionsThe sediment source assessments used in the nine TMDLsreviewed in this paper were conducted by many differentanalysts using methods that provided estimated sedimentyields with a high degree of uncertainty Even so basedupon the available data from the TMDLs reviewed here thetotal average sediment yields for the entire period of recordand the more recent (post-FPA) period show that harvestrelated sources in logged areas (ldquoin-unitsrdquo) are between 17and 18 of the total sediment production Since thepassage of the 1973 FPA the total sediment loadingresulting from all human-related activities (harvest- androad-related) decreased by only 2 (55 vs 53 for thepost-FPA period) with a small increase in natural contribu-tions (45 vs 47 for the post-FPA period) It should benoted that the harvest-related proportion varies (SE 4 vs9 for the post-FPA period) across watersheds and byindividual sediment source analysis

Although the approach used in setting sediment-basedTMDLs assumes that there is some threshold level ofincrease above natural sediment delivery that will notadversely affect salmon it is not clear what this thresholdis (ie it is not clear what levels of human-related sedimentloadings can be tolerated by coho salmon) Despite thesimilarity in the ratio of anthropogenic to natural sedimentcontributions under differing periods of analysis the total

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

NATURAL No land useassociation

72

OTHER MGMT Roadrelated

5

OTHER MGMT Skid trail related

5

HARVEST Advancedsecond growth

2

HARVEST Tractor clear cut

11

HARVEST Cable clear cut3

HARVEST Partial harvest2

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

HARVEST Mass Wasting23

HARVEST In-Unitldquosurface erosion

1

OTHER MGMT Road-related mass wasting

15

OTHER MGMT Road-related surface erosion

25

NATURAL Mass wasting 1

NATURAL Surface erosion10

NATURAL Stream bankerosion

25

Figure 8 Sediment Contributions to the Ten Mile River(1979-1999) Source Draft Ten Mile River TMDL (USEPA2000b Mathews and Associates 2000)

Figure 9 Sediment Contributions to the Van Duzen River(1955-1999) Source Van Duzen River TMDL (USEPA 1999c)

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 23

and harvest-related unit area sediment yields do appear tohave substantially decreased in the last 25 years since thepassage of the FPA Roads and skid trails continue tocontribute about 35 of the total sediment loading or two-thirds of all human-related sediment loading for bothperiods of analysis suggesting that no substantial changein road-related sediment inputs has occurred despiteimproved forest practices Considering effects of improvedforest practices on hillslope stability and erosion it isplausible that road-related mass wasting and surfaceerosion under current conditions are legacy effects of oldroads However the published sediment source analysesincluding this review analysis cannot distinguish whetherpost-FPA road-related sediment delivery originated fromolder roads or roads constructed under FPA

RecommendationsConsidering the extraordinary time and effort devoted bythe authors of the nine TMDLs reviewed here a moreconsistent approach to future sediment source analyses willallow cross-basin comparisons between individual analyses(eg the range of the coho) to answer regional scale ques-tions In order to better discriminate between individualsediment source contributions in future TMDLs we recom-mend the following

l Stratify assessments by geomorphic terrains or geologicunit type and type of land-usel Bracket aerial photo analysis and sediment sourceestimates using multiple time periods with each periodencompassing the smallest geomorphologically meaningfultime unit possible and with consideration given to naturalvariation among years in the magnitude and frequency oflarger erosion-triggering stormslmiddot Determine annual sediment yields by geomorphic processtype (eg structural landslides shallow landslidesearthflows soil creep road-related surface erosion andmass wasting and fluvial erosion on hillslopes includingsheetwash rilling and gullying)l middotDetermine annual sediment yields by managementpractice (eg skid and road-related surface erosion road-related landsliding timber harvest-related landsliding andsurface erosion vineyard and grazing-related surfaceerosion)l Future sediment budgets should focus on improvingestimates of sediment delivery ratios (ie proportion ofsediment produced on hillslopes to that delivered to streamchannel) as well as documenting changes in mean annualsediment yieldl Future sediment source analyses should distinguishbetween mass wasting and surface erosion associated withroads built under the 1973 FPA vs older roads that had nosuch controlsl Lastly future TMDL sediment source analyses shouldconsider separating out estimates of coarse vs fine sedi-ment loading especially for road-related sources of sedi-ment since the biological effects of sediment on salmon varywith sediment particle size-

AcknowledgementsWe would like to thank Bruce Orr and Frank Ligon from StillwaterSciences Mike Furniss from the Forest Service Pacific NorthwestResearch Station and Bill Dietrich from UC Berkeley for theirinput and review Steve Kramer Angela Percival Whitney Sparksand Jay Stallman provided technical support

ReferencesAhnert F 1970 Functional relationships between denudation reliefand uplift in large mid-latitude drainage basins American Journal ofScience 268 243-263Campbell RH 1975 Soil slips debris flows and rainstorms in theSanta Monica Mountains and vicinity southern California U SGeological Survey Washington D C Professional Paper No 851Dietrich WE and T Dunne 1978 Sediment budget for a smallcatchment in mountainous terrain Zeitschrift fuer Geomorphologie NF 29191-206Dietrich WE CJ Wilson and SL Reneau 1986 Hollows colluviumand landslides in soil-mantled landscapes in Hillslope ProcessesSixteenth Annual Geomorphology Symposium A Abrahams (ed) Allenand Unwin Ltd p 361-388Dietrich WE CJ Wilson DR Montgomery J McKean and RBauer 1992 Erosion thresholds and land surface morphology Geology20675-679Dietrich WE CJ Wilson DR Montgomery and J McKean 1993Analysis of erosion thresholds channel networks and landscapemorphology using a digital terrain model J Geology 101(2)161-180Dietrich W E R Real de Asua J Coyle B K Orr and M Trso1998 A validation study of the shallow slope stability modelSHALSTAB in forested lands of northern California Prepared forLouisiana-Pacific Corporation by Department of Geology and Geophys-ics University of California Berkeley and Stillwater EcosystemWatershed amp Riverine Sciences Berkeley California 29 June 1998Kelsey HM 1977 Landsliding channel changes sediment yield andland use in the Van Duzen River basin north coastal California 1941-1975 Ph D Thesis University of California Santa Cruz 370 pKeppeler E T and D Brown 1998 Subsurface drainage processesand management impacts Conference on Logging second-growthcoastal watersheds The Caspar Creek story Mendocino CommunityCollege Ukiah California California Department of Forestry and FireProtection USDA Forest Service and University of California Coopera-tive Extension 11 pagesLeeder MR 1991 Denudation vertical crustal movements andsedimentary basin infill Geol Rundsch 80441-458Madej M A 1995 Changes in channel-stored sediment RedwoodCreek northwestern California 1947 to 1980 In Geomorphic processesand aquatic habitat in the Redwood Creek basin northwesternCalifornia Edited by K M Nolan H M Kelsey and D C Marron US Geological Survey Professional Paper 1454 Washington D CMatthews amp Associates 1999 Sediment source analysis andpreliminary sediment budget for the Noyo River Weaverville CA(Contract 68-C7-0018 Work Assignment 0-18) Prepared for TetraTech Inc Fairfax VA May 1999Matthews amp Associates 2000 Sediment Source Analysis andPreliminary Sediment Budget for the Ten Mile River MendocinoCounty Ca Prepared for Tetra Tech IncMiller D J 1995 Coupling GIS with physical models to assess deep-seated landslide hazards Environmental and Engineering Geoscience1(3)263-276Montgomery D R and WE Dietrich 1994 A physically-based modelfor topographic control on shallow landsliding Water ResourcesResearch 30(4)1153-1171Montgomery D R W E Dietrich and K Sullivan 1998 The role ofGIS in watershed analysis Pages 241-261 in S N Lane K SRichards and J H Chandler editors Landform monitoring modellingand analysis John Wiley amp Sons New YorkMontgomery DR KM Schmidt HM Greenberg and WE Dietrich2000 Forest clearing and regional landsliding Geology 28311-214Pacific Watershed Associates 1997 Sediment production and deliveryin the Garcia River watershed Mendocino County California ananalysis of existing published and unpublished data Prepared forTetra Tech Inc and the US Environmental Protection AgencyPacific Watershed Associates 1999 Sediment source investigation forthe Van Duzen River watershed An aerial photo analysis and fieldinventory of erosion and sediment delivery in the 429 mi2 Van DuzenRiver Basin Humboldt County California Prepared for Tetra TechInc and the US Environmental Protection AgencyRaines M A and HM Kelsey 1991 Sediment budget for the GrouseCreek basin Humboldt County California Western WashingtonUniversity Department of Geology in cooperation with Six Rivers

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

24 WMC Networker Summer 2001

Cumulative Watershed EffectsThen and Now1

Leslie M ReidPacific Southwest Reseach Station Arcata

AbstractCumulative effects are the combined effects of multiple activi-ties and watershed effects are those which involve processesof water transport Almost all impacts are influenced bymultiple activities so almost all impacts must be evaluatedas cumulative impacts rather than as individual impactsExisting definitions suggest that to be significant an impactmust be reasonably expected to have occurred or to occur in thefuture and it must be of societally validated concern to some-one or influence their activities or options Past approaches toevaluating and managing cumulative watershed impacts havenot yet proved successful for averting these impacts so inter-est has grown in how to regulate land-use activities to reverseexisting impacts Approaches being discussed include require-ments for ldquozero net increaserdquo of sediment linkage of plannedactivities to mitigation of existing problems use of moreprotective best management practices and adoption of thresh-olds for either land-use intensity or impact level Differentkinds of cumulative impacts require different kinds of ap-proaches for management Efforts are underway to determinehow best to evaluate the potential for cumulative impacts andthus to provide a tool for preventing future impacts and fordetermining which management approaches are appropriatefor each issue in an area Future impact analysis methodsprobably will be based on strategies for watershed analysisAnalysis would need to consider areas large enough for themost important impacts to be evident to evaluate time scaleslong enough for the potential for impact accumulation to beidentified and to be interdisciplinary enough that interac-tions among diverse impact mechanisms can be understood

IntroductionTen years ago cumulative impacts were a major focus ofcontroversy and discussion Today they still are although theterm ldquoeffectsrdquo has generally replaced ldquoimpactsrdquo in part toacknowledge the fact that not all cumulative changes areundesirable However because the changes most relevant tothe issue are the undesirable ones ldquocumulative effectrdquo isusually further modified to ldquoadverse cumulative effectrdquo

The good news from the past 10 yearsrsquo record is that it was notjust the name that changed Most of the topics of discoursehave also shifted (Table 1) and this shift in focus is evidenceof some progress in understanding The bad news is thatprogress was too little to have prevented the cumulative im-pacts that occurred over the past 10 years This paper firstreviews the questions that have been resolved in order toprovide a historical context for the problem then uses ex-amples from Caspar Creek and New Zealand to examine theissues surrounding questions yet to be answered

Then What Is a Cumulative ImpactThe definition of cumulative impacts should have been atrivial problem because a legal definition already existed

National Forest and the Bureau for Faculty Research WesternWashington University Bellingham WA February 110 pagesRaines M 1998 South Fork Trinity River sediment source analysisDraft Tetra Tech Inc October 1998Redwood National and State Parks 1997 Draft Redwood CreekWatershed Analysis 81 p (March 1997)Reid LM and T Dunne 1996 Rapid evaluation of sedimentbudgets Catena Verlag Reiskirchen GermanyReneau S L and W E Dietrich 1987 Size and location of colluviallandslides in a steep forested landscape Conference on Erosion andsedimentation in the Pacific Rim Edited by R L Beschta T Blinn GE Grant F J Swanson and G G Ice IAHS Publication No 165International Association of Hydrological Scientists Corvallis OregonRoering J J J W Kirchner W E Dietrich 1999 Evidence fornonlinear diffusive sediment transport on hillslopes and implicationsfor landscape morphology Water Resources Research 35(3)853-870Sidle RC AJ Pearce CL OrsquoLoughlin 1985 Hillslope stability andland use American Geophysical Union Water Resources Monograph11 140 pSidle R C 1992 A theoretical model of the effect of timber harvestingon slope stability Water Resources Research 281897-1910Stillwater Sciences 1999 South Fork Eel TMDL Sediment SourceAnalysis Final Report Prepared for Tetra Tech Inc Fairfax VA 3August 1999Summerfield M A and N J Hulton 1994 Natural controls offluvial denudation rates in major world drainage basins Journal ofGeophysical Research 99(7) 13871-13883Swanson F J and R L Fredriksen 1982 Sediment routing andbudgets implications for judging impacts of forestry practicesWorkshop on sediment budgets and routing in forested drainagebasins proceedings Edited by F J Swanson R J Janda T Dunneand D N Swanston General Technical Report PNW-141 U S ForestService Pacific Northwest Forest and Range Experiment StationPortland OregonSwanson F J L E Benda S H Duncan G E Grant W FMegahan L M Reid and R R Ziemer 1987 Mass failures and otherprocesses of sediment production in Pacific Northwest forest land-scapes Contribution No 57 in Streamside management forestry andfishery interactions Edited by Salo E O and T W Cundy College ofForest Resources University of Washington SeattleTucker GE and R L Bras 1998 Hillslope processes drainagedensity and landscape morphology Water Resources Research34(10)2751-2764USDA 1970 Water land and related resourcesmdashnorth coastal area ofCalifornia and portions of southern Oregon Appendix 1 Sedimentyield and land treatment Eel and Mad River basins Prepared by theUSDA River Basin Planning Staff in cooperation with CaliforniaDepartment of Water Resources June 1970USEPA 1998a Garcia River Sediment Total Maximum Daily LoadUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1998b South Fork Trinity River and Hayfork Creek SedimentTotal Maximum Daily Loads US Environmental Protection AgencyRegion IX San Francisco CAUSEPA 1998c Total Maximum Daily Load for Sediment RedwoodCreek California US Environmental Protection Agency Region IXSan Francisco CA 30 December 1998USEPA 1999a Protocol for Developing Sediment TMDLs EPA 841-B-99-004 Office of Water (4503F) United States EnvironmentalProtection Agency Washington DC 132 pagesUSEPA 1999b Noyo River Total Maximum Daily Load for SedimentUS Environmental Protection Agency Region IX San Francisco CAUSEPA 1999c Van Duzen River and Yager Creek Total MaximumDaily Load for Sediment US Environmental Protection Agency RegionIX San Francisco CAUSEPA 1999d South Fork Eel River Total Maximum Daily Loads forSediment and Temperature Environmental Protection Agency RegionIX San Francisco CA December 1999USEPA 2000a Navarro River Total Maximum Daily Loads forTemperature and Sediment Public Review Draft US EnvironmentalProtection Agency Region IX San Francisco CA September 2000USEPA 2000b Ten Mile River Total Maximum Daily Load forSediment Public Review Draft US Environmental Protection AgencyRegion IX San Francisco CAWilson C J and WE Dietrich 1987 The contribution of bedrockgroundwater flow to storm runoff and high pore pressure developmentin hollows Conference on Erosion and sedimentation in the PacificRim Edited by R L Beschta T Blinn G E Grant F J Swanson

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 25

According to the Council on Environmental Quality (CEQGuidelines 40 CFR 15087 issued 23 April 1971)

ldquoCumulative impactrdquo is the impact on the environment whichresults from the incremental impact of the action when added toother past present and reasonably foreseeable future actionsregardless of what agency (Federal or non-Federal) or personundertakes such other actions Cumulative impacts can resultfrom individually minor but collectively significant actionstaking place over a period of time

This definition presented a problem though It seemed toinclude everything and a definition of a subcategory is notparticularly useful if it includes everything A lot of effort thuswent into trying to identify impacts that were modified be-cause of interactions with other impacts In particular thesearch was on for ldquosynergistic impactsrdquo in which the impactfrom a combination of activities is greater than the sum of theimpacts of the activities acting alone

In the long run though the legal definition held ldquocumulativeimpactsrdquo are generally accepted to include all impacts that areinfluenced by multiple activities or causes In essence thedefinition did not define a new type of impact Instead itexpanded the context in which the significance of any impactmust be evaluated Before regulations could be written toallow an activity to occur as long as the impacting party tookthe best economically feasible measures to reduce impacts Ifthe portion of the impact attributable to a particular activitywas not independently damaging that activity was not ac-countable Now however the best economically feasible mea-sures are no longer sufficient if the impact still occurs Theactivities that together produce the impact are responsible forthat impact even if each activity is individually responsiblefor only a small portion of the impact

A cumulative watershed impact is a cumulative impact thatinfluences or is influenced by the flow of water through awatershed Most impacts that occur away from the site of thetriggering land-use activity are cumulative watershed im-pacts because something must be transported from the activ-ity site to the impact site if the impact is to occur and water isone of the most common transport media Changes in thewater-related transport of sediment woody debris chemicalsheat flora or fauna can result in off-site cumulative water-shed impacts

Then Do Cumulative Impacts Actually OccurThe fact that the CEQ definition of cumulative impacts pre-vailed made the second question trivial almost all impactsare the product of multiple influences and activities and aretherefore cumulative impacts The answer is a resoundingldquoyesrdquo

Then How Can Cumulative Impacts Be AvoidedBecause the National Environmental Policy Act of 1969 speci-fied that cumulative effects must be considered in evaluations

of environmental impact for federal projects and permitsmethods for regulating cumulative effects had to be estab-lished even before those effects were well understood Similarlegislation soon followed in some states and private landown-ers and state regulatory agencies also found themselves inneed of approaches for addressing cumulative impacts As aresult a rich variety of methods to evaluate and regulatecumulative effects was developed The three primary strate-gies were the use of mechanistic models indices of activitylevels and analysis

Mechanistic models were developed for settings where concernfocused on a particular kind of impact On National Forests incentral Idaho for example downstream impacts of logging onsalmonids were assumed to arise primarily because of deposi-tion of fine sediments in stream gravels Abundant dataallowed relationships to be identified between logging-relatedactivities and sedimentation (Cline and others 1981) andbetween sedimentation and salmonid response (Stowell andothers 1983) Logging was then distributed to maintain lowsedimentation rates Unfortunately this approach does notaddress the other kinds of impacts that might occur and itrelies heavily on a good understanding of the locale-specificrelationships between activity and impact It cannot be ap-plied to other areas in the absence of lengthy monitoringprograms

National Forests in California initially used a mechanisticmodel that related road area to altered peak flows but themodel was soon found to be based on invalid assumptions Atthat point ldquoequivalent road acresrdquo (ERAs) began to be usedsimply as an index of management intensity instead of as amechanistic driving variable All logging-related activitieswere assigned values according to their estimated level ofimpact relative to that of a road and these values weresummed for a watershed (USDA Forest Service 1988) Furtheractivities were deferred if the sum was over the thresholdconsidered acceptable Three problems are evident with thismethod the method has not been formally tested differentkinds of impacts would have different thresholds and recoveryis evaluated according to the rate of recovery of the assumeddriving variables (eg forest cover) rather than to that of theimpact (eg channel aggradation) Cumulative impacts thuscan occur even when the index is maintained at an ldquoacceptablerdquovalue (Reid 1993)

The third approach locale-specific analysis was the methodadopted by the California Department of Forestry for use onstate and private lands (CDF 1998) This approach is poten-tially capable of addressing the full range of cumulative im-pacts that might be important in an area A standardizedimpact evaluation procedure could not be developed because ofthe wide variety of issues that might need to be assessed soanalysis methods were left to the professional judgment ofthose preparing timber harvest plans Unfortunately over-sight turned out to be a problem Plans were approved eventhough they included cumulative impact analyses that wereclearly in error In one case for example the report stated that

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

NowWhat is a ldquosignificantrdquo adverse cumulative effect

How can regulation reverse adverse cumulative effectsHow can adverse cumulative effects be avoided

ThenWhat is a cumulative impactDo cumulative impacts exist

How can cumulative impacts be avoided

Table 1mdashCommonly asked questions concerning cumulative impacts in 1988 (then) and 1998 (now)

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

26 WMC Networker Summer 2001

the planned logging would indeed introduce sediment tostreams but that downstream riparian vegetation would fil-ter out all the sediment before it did any damage Were thisactually true virtually no stream would carry suspended sedi-ment In any case even though timber harvest plans preparedfor private lands in California since 1985 contain statementsattesting that the plans will not result in increased levels ofsignificant cumulative impacts obvious cumulative impactshave accrued from carrying out those plans Bear Creek innorthwest California for example sustained 2 to 3 meters ofaggradation after the 1996-1997 storms and 85 percent of thesediment originated from the 37 percent of the watershed areathat had been logged on privately owned land during theprevious 15 years (Pacific Watershed Associates 1998)

EPArsquos recent listing of 20 north coast rivers as ldquoimpairedwaterwaysrdquo because of excessive sediment loads altered tem-perature regimes or other pervasive impacts suggests thatwhatever the methods used to prevent and reverse cumulativeimpacts on public and private lands in northwest Californiathey have not been successful At this point then we have abetter understanding of what cumulative impacts are and howthey are expressed but we as yet have no workable approachfor avoiding or managing them

The Interim ExamplesOne of the reasons that the topics of discourse have changedover the past 10 years is that a wider range of examples hasbeen studied As more is learned about how particular cumu-lative impacts develop and are expressed it becomes morepossible to predict and manage future impacts Two examplesserve here to display complementary approaches to the studyof cumulative impacts and to provide a context for discussionof the remaining questions

Studying Cumulative Impacts at Caspar Creek

Cumulative impacts result from the accumulation of multipleindividual changes One approach to the study of cumulativeimpacts therefore is to study the variety of changes caused bya land-use activity in an area and evaluate how those changesinteract This approach is essential for developing an under-standing of the changes that can generate cumulative impactsand thus for understanding the potential mechanisms of im-pact The long-term detailed hydrological studies carried outbefore and after selective logging of a second-growth redwoodforest in the 4-km2 South Fork Caspar Creek watershed andclearcut logging in 5-km2 North Fork Caspar Creek watershedprovide the kinds of information needed for this approachOther papers in these proceedings describe the variety ofstudies carried out in the area and here the results of thosestudies are reviewed as they relate to cumulative impacts

Results of the South Fork study suggest that 65-percent selec-tive logging tractor yarding and associated road managementmore than doubled the sediment yield from the catchment(Lewis these proceedings) while peak flows showed a statis-tically significant increase only for small storms near thebeginning of the storm season (Ziemer these proceedings)Sediment effects had returned to background levels within 8years of the end of logging while minor hydrologic effectspersisted for at least 12 years (Thomas 1990) Road construc-tion and logging within riparian zones has helped to perpetu-ate low levels of woody debris loading in the South Fork thatoriginally resulted from the first cycle of logging and from later

clearing of in-stream debris An initial pulse of blowdown islikely to have occurred soon after the second-cycle logging butthe resulting woody debris is now decaying Todayrsquos near-channel stands contain a high proportion of young trees andalders so debris loadings are likely to continue to decrease inthe future until riparian stands are old enough to contributewood Results of the South Fork study reflect roading loggingand yarding methods used before forest practice rules wereimplemented

Local cumulative impacts from two cycles of logging along theSouth Fork are expressed primarily in the altered channelform caused by loss of woody debris and the presence of a mainhaul road adjacent to the channel But for the presence of theSouth Fork weir pond which trapped most of the sedimentload downstream cumulative impacts could have resultedfrom the increased sediment load in combination with similarincreases from surrounding catchments Although the initialincrease in sediment load had recovered in 8 years estimatesof the time over which sediment impacts could accumulatedownstream of analogous watersheds without weirs wouldrequire information about the residence time of sediment atsites of concern downstream Recent observations suggestthat the 25-year-old logging is now contributing a second pulseof sediment as abandoned roads begin to fail (Cafferata andSpittler these proceedings) so the overall impact of logging inthe South Fork may prove to be greater than previously thought

The North Fork studies focus on the effects of clearcut logginglargely in the absence of near-stream roads The primary studywas designed to test for the presence of synergistic cumulativeimpacts on suspended sediment load and storm flows Nestedwatersheds were monitored before and after logging to deter-mine whether the magnitude of hydrologic and sediment trans-port changes increased decreased or remained constant down-stream Results showed that the short-term effects on sedi-ment load and runoff increased approximately in proportion tothe area logged above each gauging station thus suggestingthat the effect is additive for the range of storms sampledLong-term effects continue to be studied

Results also show an 89 percent increase in sediment loadafter logging of 50 percent of the watershed (Lewis theseproceedings) Peak flows greater than 4 L s-1 ha-1 which onaverage occur less than twice a year increased by 35 percent incompletely clearcut tributary watersheds although there wasno statistically significant change in peak flow at the down-stream-most gauging station (Ziemer these proceedings) Ob-servations in the North Fork watershed suggest that much ofthe increased sediment may come from stream-bank erosionheadward extension of unbuffered low-order streams andaccelerated wind-throw along buffered streams (Lewis theseproceedings) Channel disruption is likely to be caused in partby increased storm-flow volumes Increased sediment ap-peared at the North Fork weir as suspended load while bedloadtransport rates did not change significantly It is likely thatthe influx of new woody debris caused by accelerated blow-down near clearcut margins provided storage opportunities forincreased inputs of coarse sediment (Lisle and Napolitanothese proceedings) thereby offsetting the potential for down-stream cumulative impacts associated with coarse sedimentHowever accelerated blow-down immediately after loggingand selective cutting of buffer strips may have partially de-pleted the source material for future woody debris inputs (Reidand Hilton these proceedings) Bedload sediment yields may

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 27

increase if future rates of debris-dam decay and failure becomehigher than future rates of debris infall

But the North Fork of Caspar Creek drains a relatively smallwatershed It is one-tenth the size of Freshwater Creek water-shed one-hundredth the size of Redwood Creek watershedone-thousandth the size of the Trinity River watershed Inthese three cases the cumulative impacts of most concernoccurred on the main-stem channels impacts were not identi-fied as a major issue on channels the size of Caspar CreekThus though studies on the scale of those carried out atCaspar Creek are critical for identifying and understandingthe mechanisms by which impacts are generated they canrarely be used to explore how the impacts of most concern areexpressed because these watersheds are too small to includethe sites where those impacts occur Far downstream from awatershed the size of Caspar Creek doubling of suspendedsediment loads might prove to be a severe impact on watersupplies reservoir longevity or estuary biota

In addition the 36-year-long record from Caspar Creek isshort relative to the time over which many impacts are ex-pressed The in-channel impacts resulting from modificationof riparian forest stands will not be evident until residual woodhas decayed and the remaining riparian stands have regrownand equilibrated with the riparian management regime Es-tablishment of the eventual impact level may thus requireseveral hundred years

Studying Cumulative Impacts in the Waipaoa Watershed

A second approach to cumulative impact research is to workbackwards from an impact that has already occurred to deter-mine what happened and why This approach requires verydifferent research methods than those used at Caspar Creek

because the large spatial scales at which cumulative impactsbecome important prevent acquisition of detailed informationfrom throughout the area In addition time scales over whichimpacts have occurred are often very long so an understandingof existing impacts must be based on after-the-fact detectivework rather than on real-time monitoring A short-term studycarried out in the 2200-km2 Waipaoa River catchment in NewZealand provides an example of a large-scale approach to thestudy of cumulative impacts

A central focus of the Waipaoa study was to identify the long-term effects of altered forest cover in a setting with similarrock type tectonic activity topography original vegetationtype and climate as northwest California (Table 2) The majordifference between the two areas is that forest was convertedto pasture in New Zealand while in California the forests areperiodically regrown The strategy used for the study wassimilar to that of pharmaceutical experiments to identifypossible effects of low dosages administer high doses andobserve the extreme effects Results of course may depend onthe intensity of the activity and so may not be directly trans-ferable However results from such a study do give a very goodidea of the kinds of changes that might happen thus definingearly-warning signs to be alert for in less-intensively managedsystems

The impact of concern in the Waipaoa case was floodingresidents of downstream towns were tired of being flooded andthey wanted to know how to decrease the flood hazard throughwatershed restoration The activities that triggered the im-pacts occurred a century ago Between 1870 and 1900 beech-podocarp forests were converted to pasture by burning andgullies and landslides began to form on the pastures within afew years Sediment eroded from these sources began accumu-

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Sheep farmingLogging ranchingCurrent land use

Podocarp conifers southern beech hardwoods bracken scrub

Redwood Douglas fir hardwood grassland

Original vegetation

75005000Sediment yield (tkm2 -yr)

9701290Maximum elevation (m)

900 to 30001500 to 2900Rainfall (mmyr)

38ordm10rsquoS to 38ordm50rsquoS39ordm30rsquoN to 40ordm20rsquoN Latitude

21001760Area (km2)

grassland some reforestation of Monterey pines

sameCurrent vegetation

0 to 30 to 4Uplift rate (mmyr)

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Intensely sheared late Mesozoic sediments and volcanics

Tertiary sedimentary rocks

Bedrock

Waipaoa River BasinSouth Fork Eel River Basin 1Characteristic

Table 2mdashComparison of settings for the South Fork Eel River Basin and the Waipaoa River Basin

1 information primarily from Scott and Buer (1983)

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

28 WMC Networker Summer 2001

lating in downstream channels eventually decreasing channelcapacity enough that sheep farms in the valley began to floodwith every moderate storm Most of the farms had been movedto higher ground by about 1920 Today the terraces theyoriginally occupied are themselves at the level of the channelbed and 30 m of aggradation have been documented at one site(Allsop 1973) By the mid-1930rsquos aggradation had reached theWhatatutu town-site 20 km downstream forcing the entiretown to be moved onto a terrace 60 m above its originallocation

Meanwhile levees were being constructed farther downstreamand high-value infrastructure and land-use activities began toaccumulate on the newly ldquoprotectedrdquo lowlands At about thesame time as levees were constructed the frequency of severeflooding as identified from descriptions in the local newspa-per increased Climatic records show no synchronous changein rainfall patterns

The hydrologic and geomorphic changes that brought about theWaipaoarsquos problems are of the same kinds measured 10000km away in Caspar Creek runoff and erosion rates increasedwith the removal of the forest cover However the primaryreason for increased flood frequencies was not the direct effectsof hydrologic change but the indirect effects (fig 1) Had peak-flow increases due to altered evapotranspiration and intercep-tion loss after deforestation been the primary cause of flood-ing increased downstream flood frequencies would have datedfrom the 1880rsquos not from the 1910rsquos The presence of gullies inforested land down-slope of grasslands instead suggests thatthe major impact of locally increased peak-flows was thedestabilization of low-order channels which then led to down-stream channel aggradation decreased channel capacity andflooding At the same time loss of root cohesion contributed tohillslope destabilization and further accelerated aggradation

The levees themselves probably aggravated the impact nearbyby reducing the volume of flood flow that could be temporarilystored and the significance of the impact was increased by theincreased presence of vulnerable infrastructure

The impacts experienced in the Waipaoa catchment demon-strate a variety of cumulative effects First deforestation oc-curred over a wide-enough area that enough sediment couldaccumulate to cause a problem Second deforestation per-sisted allowing impacts to accumulate through time Thirdthe sediment derived from gully erosion (caused primarily byincreased local peak flows) combined with lesser amountscontributed by landslides (triggered primarily by decreasedroot cohesion) Fourth flood damage was increased by thecombined effects of decreased channel capacity increased run-off the presence of ill-placed levees and the increased presenceof structures that could be damaged by flooding (fig 1)

The Waipaoa example also illustrates the time-lags inherentnisms some level of impact was inevitable

But how much of the Waipaoa story can be used to understandimpacts in California In California although road-relatedeffects persist throughout the cutting cycle hillslopes experi-ence the effects of deforestation only for a short time duringeach cycle so only a portion of the land surface is vulnerable toexcessive damage from large storms at any given time (Ziemerand others 1991) Average erosion and runoff rates are in-creased but not by as much as in the Waipaoa watershed andpartial recovery is possible between impact cycles If as ex-pected similar trends of change (eg increased erosion andrunoff) predispose similar landscapes to similar trends ofresponse then the Waipaoa watershed provides an indication

INCREASED FLOOD DAMAGE

increased storm runoff

altered channel capacity

more floodplain settlement

increased overland flow soil

compaction

less ground-cover vegetation

more people

sheep less profitable

less rainfall interception and evapo-

transpiration

deforestation

sheep farms

less floodplain storage

levee construction aggradation

increased upland and

channel erosion less root

cohesion

Figure 1mdashFactors influencing changes in flood hazard in the Waipaoa River basin North Island New ZealandBold lines and shaded boxes indicate the likely primary mechanism of influence

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 29

of the kinds of responses that northwest California water-sheds might more gradually undergo The first response in-creased landsliding and gullying The second pervasive chan-nel aggradation Both kinds of responses are already evidentat sites in northwest California where the rate of temporarydeforestation has been particularly high (Madej and Ozaki1996 Pacific Watershed Associates 1998) suggesting thatresponse mechanisms similar to those of the Waipaoa areunderway However it is not yet known what the eventualmagnitude of the responses might be

Now What Is a ldquoSignificantrdquo Adverse CumulativeEffectThe key to the definition now lies in the word ldquosignificantrdquo andldquosignificantrdquo is one of those words that has a different defini-tion for every person who uses it Two general categories ofdefinition are particularly meaningful in this context how-ever To a scientist a ldquosignificantrdquo change is one that can bedemonstrated with a specified level of certainty For exampleif data show that there is only a probability of 013 that ameasured 1 percent increase in sediment load would appear bychance then that change is statistically significant at the 87percent confidence level irrespective of whether a 1 percentchange makes a difference to anything that anyone caresabout

The second category of definition concentrates on the nature ofthe interaction if someone cares about a change or if thechange affects their activities or options it is ldquosignificantrdquo orldquomeaningfulrdquo to them This definition does not require that thechange be definable statistically An unprecedented activitymight be expected on the basis of inference to cause significantchanges even before actual changes are statistically demon-strable Or to put it another way if cause-and-effect relation-ships are correctly understood one does not necessarily have towait for an experiment to be performed to know what theresults are likely to be and to plan accordingly

In the context of cumulative effects elements of both facets ofthe definition are obviously important According to the Guide-lines for Implementation of the California EnvironmentalQuality Act (14 CCR 15064 filed 13 July 1983 amended 27May 1997)

(g) The decision as to whether a project may have one or moresignificant effects shall be based on substantial evidence inthe record of the lead agencySubstantial evidence shall in-clude facts reasonable assumptions predicated upon factsand expert opinion supported by facts

(h) If there is disagreement among expert opinion supportedby facts over the significance of an effect on the environmentthe Lead Agency shall treat the effect as significant

In addition the following section (14 CCR 15065 filed 13 July1983 amended 27 May 1997) describes mandatory findings ofsignificance Situations in which ldquoa lead agency shall find thata project may have a significant effect on the environmentrdquoinclude those in which

(a) The project has the potential to substantially degrade thequality of the environment [or]reduce the number or re-strict the range of an endangered rare or threatened species

(d) The environmental effects of a project will cause substan-tial adverse effects on human beings either directly or indi-

rectly

ldquoSubstantialrdquo in these cases appears to mean ldquoof real worthvalue or effectrdquo Together these sections establish the rel-evance of both facets of the definition in essence a change issignificant if it is reasonably expected to have occurred or tooccur in the future and if it is of societally validated concern tosomeone or affects their activities or options ldquoSomeonerdquo inthis case can also refer to society in general the existence oflegislation concerning clean water endangered species andenvironmental quality demonstrates that impacts involvingthese issues are of recognized concern to many people TheEnvironmental Protection Agencyrsquos listing of waterways asimpaired under section 303(d) of the Clean Water Act wouldthus constitute documentation that a significant cumulativeimpact has already occurred

An approach to the definition of ldquosignificancerdquo that has beenwidely attempted is the identification of thresholds abovewhich changes are considered to be of concern Basin plansdeveloped under the Clean Water Act for example generallyadopt an objective of limiting turbidity increases to within 20percent of background levels Using this approach any studythat shows a statistically significant increase in the level ofturbidity rating curves of more than 20 percent with respect tothat measured in control watersheds would document theexistence of a significant cumulative impact Such a recordwould show the change to be both statistically meaningful andmeaningful from the point of view of what our society caresabout

Thresholds however are difficult to define The ideal thresh-old would be an easily recognized value separating significantand insignificant effects In most cases though there is noinherent point above which change is no longer benign Insteadlevels of impact form a continuum that is influenced by levelsof triggering activities incidence of triggering events such asstorms levels of sensitivity to changes and prior conditions inan area In the case of turbidity for example experiments havebeen carried out to define levels above which animals diedeath is a recognizable threshold in system response How-ever chronic impacts are experienced by the same species atlevels several orders of magnitude below these lethal concen-trations (Lloyd 1987) and there is likely to be an incrementaldecrease in long-term fitness and survival with each incre-ment of increased turbidity In many cases the full implica-tions of such impacts may be expressed only in the face of anuncommon event such as a drought or a local outbreak ofdisease Any effort to define a meaningful threshold in such asituation would be defeated by the lack of information concern-ing the long-term effects of low levels of exposure

If a threshold cannot be defined objectively on the basis ofsystem behavior or impact response the threshold would needto be identified on the basis of subjective considerationsDefinition of subjective thresholds is a political decision re-quiring value-laden weighting of the interests of those produc-ing the impacts and those experiencing the impacts

It is important to note that an activity is partially responsiblefor a significant cumulative impact if it contributes an incre-mental addition to an already significant cumulative impactFor example if enough excess sediment has already beenadded to a channel system to cause a significant impact thenany further addition of sediment also constitutes a significantcumulative impact

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

30 WMC Networker Summer 2001

Now How Can Regulation Reverse AdverseCumulative EffectsIn Californiarsquos north coast watersheds the prevalence ofstreams listed as impaired under section 303(d) of the CleanWater Act demonstrates that significant cumulative impactsare widespread in the area Forest management grazing andother activities continue in these watersheds so the focus ofconcern now is on how to regulate management of these landsin such a way as to reverse the impacts Current regulatorystrategies largely reflect the strategies for assessing cumula-tive impacts that were in place 10 years ago and the need forchanging these regulatory strategies is now apparent Ap-proaches to regulation that are being discussed include attain-ment of ldquozero net increaserdquo in sediment offsetting of impactsby mitigation adoption of more stringent standards for spe-cific land-use activities and use of threshold-based methods

The ldquozero-net-increaserdquo approach is based on an assumptionthat no harm is done if an activity does not increase the overalllevel of impact in an area This is the approach instituted toregulate sediment input in Grass Valley Creek in the TrinityBasin where erosion rates are to be held at or below the levelspresent in 1986 (Komar 1992) Unfortunately this approachcannot be used to reverse the trend of impacts already occur-ring because the existing trend of impact was created by thelevels of sediment input present in 1986 To reverse impactsinputs would need to be decreased to below the levels of inputthat originally caused the problem

ldquoZero-net increaserdquo requirements are often linked to mitiga-tion plans whereby expected increases in sediment productiondue to a planned project are to be offset by measures institutedto curtail erosion from other sources Some such plans evenprovide for net decreases in sediment production in a water-shed Unfortunately this approach also falls short of reversingexisting impacts because mitigation measures usually aredesigned to repair the unforeseen problems caused by pastactivities It is reasonable to assume that the present planswill also result in a full complement of unforeseen problemsbut the possibility of mistakes is generally not accounted forwhen likely input rates from the planned activities are calcu-lated Later when the unforeseen impacts become obviousrepair of the new problems would be used as mitigation forfuture projects To ensure that such a system does more thanperpetuate the existing problems it would be necessary torequire that all future impacts from a plan (and its associatedroads) are repaired as part of the plan not as mitigationmeasures to offset the impacts of future plans

In addition offsetting mitigation activities are usually ac-counted for as though the predicted impacts were certain tooccur if those activities are not carried out In reality there isonly a small chance that any given site will fail in a 5-yearperiod Appropriate mitigation would thus require that con-siderably more sites be repaired than are ordinarily allowedfor in mitigation-based plans Furthermore mitigation at onesite does not necessarily offset the kind of impacts that willaccrue from a planned project If the project is located whereimpacts from a given sediment input might be particularlysevere offsetting measures in a less-sensitive area would notbe equivalent Similarly mitigation of one kind of source doesnot cancel the impact of another kind of source Mitigationcapable of offsetting the impacts from construction of a newroad would need to include obliteration of an equal length of oldroad to offset hydrologic changes as well as measures to offset

short-term sediment inputs from construction and oblitera-tion and long-term inputs from future road use

The timing of the resulting changes may also negate theeffectiveness of mitigation measures If a project adds tocurrent sediment loads in a sediment-impaired waterwaywhile the mitigation work is designed to decrease sedimentloads at some time in the future (when the repaired sites wouldotherwise have failed) the plan is still contributing to asignificant cumulative impact irrespective of the offsettingmitigation activities In other words if a watershed is alreadyexperiencing a significant sediment problem it makes littlesense to use an as-yet-unfulfilled expectation of future im-provement as an excuse to make the situation worse in theshort term It would thus be necessary to carry out mitigationactivities well in advance of the activities which they aredesigned to offset so that impact levels are demonstrablydecreasing by the time the unavoidable new impacts aregenerated

The third approach to managing existing impacts is the adop-tion of more stringent standards that are based on the needsof the impacted resources Attempts to avert cumulative im-pacts through the implementation of ldquobest management prac-ticesrdquo (BMPs) have failed in the past in part because they werebased strongly on the economic needs of the impacting landuses and thus did not fully reflect the possibility that signifi-cant adverse cumulative effects might accrue even from re-duced levels of impact A new approach to BMPs has recentlyappeared in the form of standards and guidelines for designingand managing riparian reserves on federal lands affected bythe Northwest Forest Plan (USDA and USDI 1994) Guide-lines for the design of riparian reserves are based on studiesthat describe the distance from a forest edge over which themicroclimatic and physical effects of the edge are evident andhave a principal goal of producing riparian buffer strips ca-pable of adequately shielding the aquatic systemmdashand par-ticularly anadromous salmonidsmdashfrom the effects of upslopeactivities Any land-use activities to be carried out within thereserves must be shown not to incur impacts on the aquaticsystem Even with this level of protection the NorthwestForest Plan is careful to point out that riparian reserves andtheir accompanying standards and guidelines are not in them-selves sufficient to reverse the trend of aquatic habitat degra-dation These measures are expected to be effective only incombination with (1) watershed analysis to identify the causesof problems (2) restoration programs to reverse those causesand speed recovery and (3) careful protection of key water-sheds to ensure that watershed-scale refugia are present TheNorthwest Forest Plan thus recognizes that BMPs alone arenot sufficient although they can be an important component ofa broader landscape-scale approach to recovery from impacts

The final approach is the use of thresholds Threshold-basedmethods would allow for altering land-use prescriptions oncea threshold of concern has been surpassed This in essence isthe approach used on National Forests in California if theindex of land-use intensity rises above a defined thresholdvalue further activities are deferred until the value for thewatershed is once again below threshold Such an approachwould be workable if there is a sound basis for identifyingappropriate levels of land-use intensity This basis wouldneed to account for the occurrence of large storms becauseactual impact levels rarely can be identified in the absence ofa triggering event The approach would also need to includeprovisions for frequent review so that plans could be modified

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 31

if unforeseen impacts occur

Thresholds are more commonly considered from the point ofview of the impacted resource In this case activities arecurtailed if the level of impact rises above a predeterminedvalue This approach has limited utility if the intent is toreverse existing or prevent future cumulative impacts becausemost responses of interest lag behind the land-use activitiesthat generate them If the threshold is defined according tosystem response the trend of change may be irreversible bythe time the threshold is surpassed In the Waipaoa case forexample if a threshold were defined according to a level ofaggradation at a downstream site the system would havealready changed irreversibly by the time the effect was visibleThe intolerable rate of aggradation that Whatatutu experi-enced in the mid-1930rsquos was caused by deforestation 50 yearsearlier Similarly the current pulse of aggradation near themouth of Redwood Creek was triggered by a major storm thatoccurred more than 30 years ago (Madej and Ozaki 1996) Incontrast turbidity responds quickly to sediment inputs butrecognition of whether increases in turbidity are above a thresh-old level requires a sequence of measurements over time toidentify the relation between turbidity and discharge and itrequires comparison to similar measurements from an undis-turbed or less-disturbed watershed to establish the thresholdrelationship

The potential effectiveness of the strategies described abovecan be assessed by evaluating their likely utility for address-ing particular impacts (table 3) In North Fork Caspar Creekfor example suspended sediment load nearly doubled afterclearcutting with the change largely attributable to increasedsediment transport in the smallest tributaries because ofincreased runoff and peakflows Strategies of zero-net-increaseand offsetting mitigations would not have prevented the changebecause the effect was an indirect result of the volume ofcanopy removed hydrologic change is not readily mitigableBMPs would not have worked because the problem was causedby the loss of canopy not by how the trees were removedImpacts were evident only after logging was completed soimpact-based thresholds would not have been passed untilafter the change was irreversible Only activity-based thresh-olds would have been effective in this case because hydrologicchange is roughly proportional to the area logged the magni-tude of hydrologic change could have been managed by regulat-ing the amount of land logged

A second long-term cumulative impact at Caspar Creek is thechange in channel form that is likely to result from pastpresent and future modifications of near-stream forest standsIn this case a zero-net-change strategy would not have workedbecause the characteristics for which change is of concernmdash

debris loading in the channelmdashwill be changing to an unknownextent over the next decades and centuries in indirect responseto the land-use activities Off-setting mitigations would mostlikely take the form of artificially adding wood but such ashort-term remedy is not a valid solution to a problem thatmay persist for centuries In this case BMPs in the form ofriparian buffer strips designed to maintain appropriate de-bris infall rates would have been effective Impact thresholdswould not have prevented impacts as the nature of the impactwill not be fully evident for decades or centuries Activitythresholds also would not be effective because the recoveryrate of the impact is an order of magnitude longer than thelikely cutting cycle

The Waipaoa problem would also be poorly served by most ofthe available strategies Once underway impacts in theWaipaoa watershed could not have been reversed throughadoption of zero-net-increase rules because the importance ofearlier impacts was growing exponentially as existing sourcesenlarged Similarly mitigation measures to repair existingproblems would not have been successful the only effectivemitigation would have been to reforest an equivalent portionof the landscape thus defeating the purpose of the vegetationconversion BMPs would not have been effective because howthe watershed was deforested made no difference to the sever-ity of the impact Thresholds defined on the basis of impactalso would have been useless because the trend of change waseffectively irreversible by the time the impacts were visibledownstream Thresholds of land-use intensity however mighthave been effective had they been instituted in time If only aportion of the watershed had been deforested hydrologic changemight have been kept at a low enough level that gullies wouldnot have formed The only potentially effective approach in thiscase thus would have been one that required an understandingof how the impacts were likely to come about De facto institu-tion of land-use-intensity thresholds is the approach that hasnow been adopted in the Waipaoa basin to reduce existingcumulative impacts The New Zealand government bought themajor problem areas and reforested them in the 1960rsquos and1970rsquos Over the past 30 years the rate of sediment input hasdecreased significantly and excess sediment is beginning tomove out of upstream channels

It is evident that no one strategy can be used effectively tomanage all kinds of cumulative impacts To select an appro-priate management strategy it is necessary to determine thecause symptoms and persistence of the impacts of concernOnce these characteristics are understood each availablestrategy can be evaluated to determine whether it will havethe desired effect

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

no

no

no

Off-settingmitigations

no

no

no

Zero-net-increase

no

YES

no

Impact-based Best Management Practices

YESnoWaipaoa flooding from channel aggradation

YESnoCaspar Creek sediment yield increase from hydrologic change

nonoCaspar Creek channel change from altered wood regime

ActivityThresholds

Impact Threshold

Cumulative Impact

Table 3mdashPotential effectiveness of various strategies for managing specific cumulative impacts

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

32 WMC Networker Summer 2001

Now How Can Adverse Cumulative Effects BeAvoidedThe first problem in planning land use to avoid cumulativeeffects is to identify the cumulative effects that might occurfrom a proposed activity A variety of methods for doing so havebeen developed over the past 10 years and the most widelyadopted of these are methods of watershed analysis Washing-ton State has developed and implemented a procedure todesign management practices to fit conditions within specificwatersheds (WFPB 1995) with the intent of holding futureimpacts to low levels A procedure has also been developed forevaluating existing and potential environmental impacts onfederal lands in the Pacific Northwest (Regional EcosystemOffice 1995) Both methods have strengths and weaknesses

The Washington approach describes detailed methods forevaluating processes such as landsliding and road-surfaceerosion and provides for participation of a variety of interestgroups in the analysis procedure Because the approach wasdeveloped through consensus among diverse groups it is widelyaccepted However methods have not been adequately testedand the approach is designed to consider only issues related toanadromous fish and water quality In general only thoseimpacts which are already evident in the watershed are usedas a basis for invoking prescriptions more rigorous than stan-dard practices No evaluation need be done of the potentialeffects of future activities in the watershed it is assumed thatthe activities will not produce significant impacts if the pre-scribed practices are followed The method does not evaluatethe cumulative impacts that might result from implementa-tion of the prescribed practices and does not provide for evalu-ating the potential of future activities to contribute to signifi-cant cumulative impacts Collins and Pess (1997a 1997b)provide a comprehensive review of the approach

The Federal interagency watershed analysis method in con-trast was intended simply to provide an interdisciplinarybackground understanding of the mechanisms for existing andpotential impacts in a watershed The Federal approach recog-nizes that which activities are appropriate in the future willdepend on watershed conditions present in the future so thatcumulative effects analyses would still need to be carried outfor future activities Although the analyses were intended to becarried out with close interdisciplinary cooperation analyseshave tended to be prepared as a series of mono-disciplinarychapters

Both procedures suffer from a tendency to focus exclusively onareas upstream of the major impacts of concern The potentialfor cumulative effects cannot be evaluated if the broadercontext for the impacts is not examined To do so an area largeenough to display those impacts must be examined Becauseof Californiarsquos topography and geography the most importantareas for impact are at the mouths of the river basins that iswhere most people live where they obtain their water whereall anadromous fish must pass if they are to make their wayupstream and where the major transportation routes crossThese are also sites where sediment is likely to accumulateThe Washington method considers watersheds smaller than200 km2 and the Federal Interagency method evaluates wa-tersheds of 50 to 500 km2 neither method consistently in-cludes the downstream areas of most concern

In addition a broad enough time scale must be evaluated if thepotential for accumulation of impacts is to be recognized Inthe Waipaoa case for example impacts were relatively minorduring the years immediately following deforestation aggra-

dation was not evident until after a major storm had occurredIn the South Fork of Caspar Creek the influences of logging onsediment yield and runoff were thought to have largely disap-peared within a decade However during the 1997-98 winterthree decades after road construction destabilization of oldroads has led to an increase in landslide frequency (Cafferataand Spittler these proceedings) It is possible that a majorsediment-related impact from the past land-use activities isyet to come In any case the success of a land-use activity inavoiding impacts is not fully tested until the occurrence of avery wet winter a major storm a protracted drought and otherraremdashbut expectedmdashevents

Of particular concern in the evaluation of future impacts is thepotential for interactions between different mechanisms ofchange In the Waipaoa case for example hydrologic changescontributed to a severe increase in flood hazard less because oftheir direct influence on downstream peak-flow dischargesthan because they accelerated erosion thus leading to aggra-dation and decreased channel capacity In retrospect such achange is clearly visible in prospect it would be difficult toanticipate In other cases unrelated changes combine to aggra-vate a particular impact Over-winter survival of coho salmonmay be decreased by simplification of in-stream habitat due toincreased sediment loading at the same time that access todownstream off-channel refuges is blocked by construction offloodplain roads and levees The overall effect might be a severedecrease in out-migrants whereas if only one of the impactshad occurred populations might have partially compensatedfor the change by using the remaining habitat option moreheavily In both of these cases the implications of changesmight best be recognized by evaluating impacts from the pointof view of the impacted resource rather than from the point ofview of the impacting land use Such an approach allowsconsideration of the variety of influences present throughoutthe time frame and area important to the impacted entityAnalysis would then automatically consider interactions be-tween the activity of interest and other influences rather thanfocusing implicitly on the direct influence of the activity inquestion

The overall importance of an environmental change can beinterpreted only relative to an unchanged state In areas aspervasively altered as northwest California and New Zealandexamples of unchanged sites are few Three strategies can beused to estimate levels of change in such a situation Firstoriginal conditions can be inferred from the nature of existingconditions and influences No road-related sediment sourceswould have been present under natural conditions for ex-ample and the influence of modified riparian stand composi-tion on woody debris inputs can be readily estimated Secondless disturbed sites can be compared with more disturbed sitesto identify the trend of change even if the end point of ldquoundis-turbedrdquo is not present Third information from analogousundisturbed sites elsewhere can often be used to provide anestimate of undisturbed conditions if it can be shown thatthose sites are similar enough to the area in question to bereasonable analogs

Once existing impacts are recognized and their causal mecha-nisms understood the potential influence of planned activitiescan be inferred from an understanding of how those activitiesmight influence the causal mechanisms

Neither of the widely used watershed analysis methods pro-vides an adequate assessment of likely cumulative effects ofplanned projects and neither makes consistent use of a varietyof methods that might be used to do so However both ap-

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

WMC Networker Summer 2001 33

collaborative learning system on the dynamics ecology andhuman interaction with watersheds the underlying frame-work for organizing the ideas resources and people involvedtranscends the subject matter Rather than confine knowledgeand learning about watersheds within boundaries this projectprovides open doorways to link in new information and learnerexperts

What is the opportunityA decade ago if somebody talked about ldquohyperlinking to-

ward consiliencerdquo few listeners would have had a clue what itwas about Common experience since the advent of televisionin the 1950s and the World Wide Web since 1994 has rein-forced a collective notion that instant graphically realisticinformation about any subject can be made available and thatit might be possible to tie it all together to make sense movingtoward consilience The soil has been prepared for the Con-tinuum Project

Electronic documents and other digital educational materi-als need to be organized by context A rich array of resourcesabout watershed ecology and management including peoplewith great wisdom and knowledge is available freely in theworld poised to be made available via computers and theInternet Texts and research that has lain dormant for decadescan be displayed on a screen at the press of a button Digitalvideo can take us places and show us things with a level ofrealism and detail that while not quite the same as a rainy-day climb into the canopy of an old-growth forest is morecomfortable and probably more accessible But this mass ofmaterial is organized in chunks rather than as a whole Thework of reconciling the knowledge and efforts of multipledisciplines remains to be done The map of key concepts in therealm of the watershed needs to be overlaid on the knowledgebase

New multimedia technologies make content more acces-sible The advent of broadband communications digital videoand streaming content mean that the richness and absolutevolume of information that can be shared is far greater thanever before An expert explanation on video coupled withsupporting scientific papers rich sources of data collected overtime and opportunities for interactive dialog can all be co-located in the same virtual space linked by place and concep-tual context

New content organization and delivery systems are avail-able Standards are maturing for structuring informationresources and human interactions so they may be catalogedsearched and shared even though the individual componentsand participants are in many different locationsContinues on Page 38

The ContinuumContinued from Page 3

proaches are instructive in their call for interdisciplinaryanalysis and their recognition that process interactions mustbe evaluated over large areas if their significance is to beunderstood At this point it should be possible to learn enoughfrom the record of completed analyses to design a watershedanalysis approach that will provide the kinds of informationnecessary to evaluate cumulative impacts and thus to under-stand specific systems well enough to plan land-use activitiesto prevent future impacts

ConclusionsUnderstanding of cumulative watershed impacts has increasedgreatly in the past 10 years but the remaining problems aredifficult ones Existing impacts must be evaluated so thatcausal mechanisms are understood well enough that they canbe reversed and regulatory strategies must be modified tofacilitate the recovery of damaged systems Methods imple-mented to date have fallen short of this goal but growingconcern over existing cumulative impacts suggests that changeswill need to be made soon-

ReferencesAllsop F 1973 The story of Mangatu Wellington New ZealandGovernment PrinterCDF [California Department of Forestry and Fire Protection] 1998Forest practice rules Sacramento CA California Department ofForestry and Fire ProtectionCline R Cole G Megahan W Patten R Potyondy J 1981 Guidefor predicting sediment yields from forested watersheds [MissoulaMT] Northern Region and Intermountain Region Forest ServiceUS Department of AgricultureCollins Brian D Pess George R 1997a Evaluation of forest practicesprescriptions from Washingtonrsquos watershed analysis program Jour-nal of the American Water Resources Association 33(5) 969-996Collins Brian D Pess George R 1997b Critique of Washingtonrsquoswatershed analysis program Journal of the American Water Re-sources Association 33(5) 997-1010Komar James 1992 Zero net increase a new goal in timberlandmanagement on granitic soils In Sommarstrom Sari editor Proceed-ings of the conference on decomposed granitic soils problems andsolutions October 21-23 1992 Redding CA Davis CA UniversityExtension University of California 84-91Lloyd DS 1987 Turbidity as a water quality standard for salmonidhabitats in Alaska North American Journal of Fisheries Manage-ment 7(1) 34-45Madej MA Ozaki V 1996 Channel response to sediment wavepropagation and movement Redwood Creek California USA EarthSurface Processes and Landforms 21 911-927Pacific Watershed Associates 1998 Sediment source investigationand sediment reduction plan for the Bear Creek watershed HumboldtCounty California Report prepared for the Pacific Lumber CompanyScotia CaliforniaRegional Ecosystem Office 1995 Ecosystem analysis at the water-shed scale version 22 Portland OR Regional Ecosystem Office USGovernment Printing Office 1995-689-12021215 Region no 10Reid LM 1993 Research and cumulative watershed effects GenTech Rep PSW-GTR-141 Berkeley CA Pacific Southwest ResearchStation Forest Service US Department of AgricultureScott Ralph G Buer Koll Y 1983 South Fork Eel Watershed ErosionInvestigation California Department of Water Resources NorthernDistrictStowell R Espinosa A Bjornn TC Platts WS Burns DCIrving JS 1983 Guide to predicting salmonid response to sedimentyields in Idaho Batholith watersheds [Missoula MT] NorthernRegion and Intermountain Region Forest Service US Departmentof AgricultureThomas Robert B 1990 Problems in determining the return of awatershed to pretreatment conditions techniques applied to a study atCaspar Creek California Water Resources Research 26(9) 2079-2087USDA and USDI [United States Department of Agriculture andUnited States Department of the Interior] 1994 Record of Decision foramendments to Forest Service and Bureau of Land Managementplanning documents within the range of the northern spotted owlStandards and Guidelines for management of habitat for late-succes-sional and old-growth forest related species within the range of thenorthern spotted owl US Government Printing Office 1994 - 589-

11100001 Region no 10USDA Forest Service [US Department of Agriculture Forest Ser-vice] 1988 Cumulative off-site watershed effects analysis ForestService Handbook 250922 Soil and water conservation handbookSan Francisco CA Region 5 Forest Service US Department ofAgricultureWashington Forest Practices Board 1995 Standard methodology forconducting watershed analysis under Chapter 222-22 WAC Version30 Olympia WA Forest Practices Division Department of NaturalResourcesZiemer RR Lewis J Rice RM Lisle TE 1991 Modeling thecumulative watershed effects of forest management strategies Jour-nal of Environmental Quality 20(1) 36-421 Originally published in Ziemer Robert R technical coordinator 1998

Gen Tech Rep PSW-GTR-168 Albany CA Pacific13Southwest Research StationForest Service US Department of Agriculture 149 p httpwwwpswfsfedusTech_PubDocumentsgtr-168gtr168-tochtml

Proceedings of the conference on coastal watersheds the Caspar Creek Story 6 May 1998

34 WMC Networker Summer 2001

Cumulative Watershed EffectsStatus Gaps and NeedsSeptember 7-8 2000 Sacramento

On September 7 and 8 2000 over 150 people gathered inSacramento to discuss cumulative watershed effects (CWE)at a conference entitled Cumulative Watershed Effects Sta-tus Gaps and Needs Specifics on the agenda and ldquoproceed-ingsrdquo from the conference and the conference organizing groupare available at httpwwwcnrberkeleyeduforestrywatershedhtml This article summarizes aspects of the con-ference

The stated goal of the conference was tomdashl Further the dialogue on developing better ways to assessand manage Cumulative Watershed Effects especially inforested mixed-ownership watersheds in California

Conference objectives includedmdashl Summarize current issues and information on CWEassessment and managementl Exchange perspectives on CWE assessment and manage-ment and identify similarities and differences among phi-losophies and approaches to CWE assessmentl Identify logical next steps for advancing CWE assessmentand management

Desired outcomes weremdashl Post-conference proceedings including Resource Guidel Agreement among participants on next steps (eg form aworking group to continue the dialogue by addressing spe-cific policy and technical issues)l Agreement on conceptual framework on assessing CWE

Was the conference goal achieved In a word yes Thisconference re-affirmed the continuing diversity of opinions onCWE assessment and offered promise for improved approachesto address the topic However no silver bullets for managingCWE were identified

Achievement of the objectives and desired outcomes was lesseasy to assess A variety of presentations were made onrelevant topics by academics agency and industry representa-tives and the public This was not a ldquohow tordquo meeting ratherthe focus was on steps to take to address CWE ldquoConceptualframeworksrdquo and multiple-step procedures were described asguidance for CWE assessment These frameworks differed indetail but participants affirmed the general utility of a concep-tual framework coupled with a toolbox of techniques to ad-dress more specific CWE issues

A variety of ldquonext stepsrdquo were proposed Partly because theconference was not aimed at tools and techniques partici-pants advocated discussion of specific assessment methodsRequests in this area included

l ldquoWorkshop discussing technical approaches to quantifydifferent types of uncertainty in watershed analysis andCWErdquol ldquo how to predict CErsquos and more importantly to predictthresholds of concernrdquol ldquoWhat about tackling some of the basic and fundamentalmeasurements we need to do CWEs Gauging stationssediment measurements rdquol ldquoHow do we do watershed analysisCWE analysis in realon-the-ground itemsrdquo

Other requests werel ldquo[to] develop a multi-disciplinary team to develop aCWE process and conduct pilotrdquol ldquo to talk more openly about the ETHICS Less abouttechnical aspectsrdquol ldquo address CWE analysis in urban and mixed-land usewatershedsrdquol ldquo provide some kind of funding for technical assistancefor residence or stakeholders in watershed analysis rdquo

Since the conference the organizing group continues to meetand has morphed into an informal action committee Thisgroup is exploring options to formalize an interagency workgroup to tackle next steps including the possible sponsorshipof a pilot demonstration project for interagency CWE assess-ment Notes from meetings of that group will be available onan expanded version of the ldquoconferencerdquo website

Specific products from the conference include a Resource Guideincorporating a listing of relevant web sites an extensivebibliography and a networking list of participants their con-tact information and their area of CWE interest and expertiseAbstracts andor multi-page texts are given for most presenta-tions This conference was unusual in capturing the real-timeessence of the meeting through artwork Words and ideas werevisually encapsulated providing focus and a level playing fieldfor all participants The artwork on the web site replicates thereal-time proceedings

CWE IssuesDiscussions identified a broad range of related issues thattogether add complexity (whatrsquos new) to assessment of CWEThese issues include

l Relationship between CWE assessment and other analy-ses (eg watershed roads watershed condition)l Multi-scale considerations in both time and spacel Data inadequaciesmdashvariability accuracy areal coverage(or the flip-side too much data)l Interpretation inadequaciesl Who does the analysis (eg is the small landowner theappropriate entity to conduct a CWE assessment)l Collaborationstakeholder involvementl Decision making and uncertaintyl Conflicting beneficial usesl Onsiteoffsite recovery ratesl Uncertainty re past disturbances and routing of water-shed productsl (In)adequacy of BMPs

Many of these issues were concretely described in a paneldiscussion on Freshwater Creek In particular uncertaintiesabout determining the true implications of an action espe-cially in light of potential legacy effects from previous actionswere questioned

Conceptual FrameworkOne approach for dealing with many of these issues is tosystematically follow an agreed-upon sequence of steps forCWE assessment One candidate sequence was offered by LeeMacDonald This ldquoconceptual frameworkrdquo is similar to othersproposed for related assessments and includes scoping analy-sis and management phases in an iterative processes thatfeeds back to earlier steps (as needed)

The scoping phase shouldmdash

WMC Networker Summer 2001 35

l Identify issues and resources of primary concernl Define the temporal and spatial scale for the assessmentl Identify the relative magnitude of risk to each resourcel Select the appropriate level of effort for the assessment

The analysis phase shouldmdashl Identify key cause-and-effect mechanismsl Estimate the range of natural variability and relativecondition for the resource(s) of concernl Identify past present and future activitiesl Evaluate the relative impact of past present and expectedfuture activitiesl Evaluate the validity and sensitivity of the predicted CE

The management phase shouldmdashl Identify possibilities for modification mitigation plan-ning and restorationl Identify key data gaps and monitoring needs

Going in the Right DirectionOur institutions are accelerating efforts to address water-shed-scale issues Watershed information systems are nowbeginning to appear and the state is funding the establish-ment of a data clearinghouse for the north coast Analyticaltools that combine in a single system data collected from amyriad of sources are coming on line and offer the promise forassessment of disparate conditions Dramatic funding in-creases at the state level are evident in increased staffing ofRegional Boards and for landowner incentives Regulatorygaps are being filled and local solutions are occurring ldquoAll-party monitoringrdquo and other collaborative approaches can bemechanisms for successfully incorporating diverse perspec-tives and assuring that local projects will be responsive toconcerns of diverse stakeholders

In summary in the 30 years since NEPA and CEQA werecrafted we have figured out when cumulative impacts assess-ments are required and have established procedures for re-view with a paperwork trail to satisfy legal requirements inthe record of decision But we are still searching for appropri-ate analysis methods to measure and predict cumulativeimpacts This doesnrsquot mean that we havenrsquot tried CWE andwatershed analyses are not panaceas science and policy areiterative processes that have advanced since NEPA and CEQAwere enacted

The conference website is the place to go to fill in the flesh forthe topics only touched on in this summary Two other relevantdocuments are the 1999 publicationsmdash

l ldquoReport of the Scientific Review Panel on CaliforniaForest Practice Rules and Salmonid Habitatrdquo httpwwwcerescagovcradownload_srphtml and

l ldquoCumulative Impacts Analysis - A Report of CDF DirectorrsquosTHP Task Forcerdquo httpfrapcdfcagovpublicationscuminpdf

US FOREST SERVICECALIFORNIA REGIONin cooperation withCALIFORNIA DEPARTMENT OF CONSERVATION

THE LITTLE HILLSIDE

written bySixth Grade Pupils of Sherwood School Salinas CA

May G Moon Teacher

with pictures drawn byFred R Holland Art Supervisor

San Bernardino City Schools

Edited and arranged byCharles E Fox US Forest Service

Edward F Dolder California Dept of Natural Resources

Original Publication byCalifornia Dept of Conservation

Sacramento CA June 1964

A nation deprived of its liberty may winit a nation divided may reunite but a nationwhose natural resources are destroyed mustinevitably pay the penalty of poverty degra-dation and decay

- Gifford Pinchot

36 WMC Networker Summer 2001

The little bare hillside laythere--the hot sun scorchingit in the summer Rain beatdown on its face in thewinter washing away thebits of dust that might havesettled there On cold winternights raindrops huddledtogether in a crack or ahollow there They froze intochunks of ice--ice thatswelled up and pried openthe cracks a little wider

It was a lonely little hillside--no grass noflowers no trees You see there was no soil--nosoil to hold the seeds no soil for their roots andno soil to hold water for plants to drink

Yes the little hillside was verylonely Then one day the windblew some seeds--little seeds that fellinto the cracks and because theywere little seeds the rain couldnrsquotfind them to washthem away

When winter came and the plants went to sleepthe mother plants made a bed for the baby seedsand held them close When the rains came thestems and leaves and roots of the baby plantseven though they were small held a little of therain Then when they had more food and waterthey grew larger and had more baby seeds Everyyear they grew larger and stronger and theirroots went down farther

One day the wind broughtsome cottonwood seeds flyingover the hills Because theplants had made a good placefor seeds to rest thecottonwood seeds stoppedand decided to stay

The next year little cottonwood treesstarted growing The little hillside feltmuch better now It was not lonely anymore

When the warm rains came in the spring thelittle seeds grew into plants The plants didnrsquothave much food and they didnrsquot have much water

so they didnrsquot grow very bigbut they kept growing andfinally had seeds of their own

The birds flying over stoppedto rest and to eat berries andseeds They dropped some to theground The grass and leavesheld them warm during thewinter When the sun shonewarm in the spring more plantsand berries grew and the littlehillside was very happy

WMC Networker Summer 2001 37

The stems and leaves and roots of the plantsheld the rain The rain stayed on the hillsideinstead of flowing away All summer a littlestream of water trickled down the hillside givinga drink to the thirsty grass and flowers and treeswhen the sun was hot

The birds told the animals about the lovelyhillside and the cool stream and the pleasantshade Many animals came to see--mother deerwith their baby fawns little squirrels and rabbitsand animals that dug homes in the soil andstored their food Fish came to live in the coolstream

The leaves fell andthey made more richsoil Roots held the soiland kept it fromwashing away Thelittle hillside and all ofits friends were happy

Then people found the little valley by thelittle hillside with the flowers and thestream and the fish and the deer Most ofthe people were very thoughtful Theyenjoyed the shade and the flowers and thecool water They ate their lunches on thegreen grass beside the cool stream They

picked up their trash and papersand left the little valley as clean asthey had found it for others toenjoy

But one day somecareless people cameThey made a fire and

had a picnic When they were readyto go home they left their plates andboxes but it didnrsquot really matter verymuch because they left their fireburning too

A little whirlwind came racingdown the hillside He blew the fireover to some leaves--dry leaves-- andthey started to burn The harder thelittle wind blew trying to put the fireout the faster it burned

The next day the little hillside wasvery sad The trees were gone thegrass and the flowers were gone Allof the birds and animals were goneThe little hillside knew the sunwould beat down hot in the summer

One day some boys and girls saw

38 WMC Networker Summer 2001

What problem do we intend to solveThe tendency of the physical and social sciences and the

humanities to break knowledge into its component parts andto focus intently in areas of specialization has led to tremen-dous advances in understanding ndash of parts at the expense ofthe whole The essential problem the Continuum Project ad-dresses is the subject of the ancient Parable of the Blind Menand the Elephanthellip each ldquosawrdquo the elephant based on theirlimited experiences To understand watersheds and to betterunderstand the different perspectives on their managementwe need to connect knowledge across disciplines

Professionals need incentives to continue learning tobroaden their knowledge continually and to provide leader-ship in forming a better collective understanding of water-sheds

Whether it is due to lack of time or competition for atten-tion most people donrsquot read much Most professionals engagedin watershed management canrsquot keep up with the volume ofavailable research and literature A method for rapidly andeffectively disseminating information among professionals isneeded that is engaging enough to attract a larger percentageof the audience than printed materials do now

Society now expects Land management policy decisions tobe effectively informed by applicable scientific findings andthat land managers will take scientific approaches to thecomplex decisions they make and implement To be moreeffective in this experts are needed to filter and synthesize thelarge body of literature available selecting adapting andinterpreting it for land managers

How do we intend to solve itBy mapping key concepts in watershed ecology and linking

them within and across the disciplines they involve the Con-tinuum Project provides a visually exciting multiple-entryresource for professional development It is comfortable forself-paced learners and provides a resource for teachers devel-oping structured outcome-based instruction

Just as modern maps are continually evolving documentswith multiple layers the ldquomappingrdquo of concepts vocabulariesand the relationships between things in the Continuum Projectis open and changeable The overall ldquoontologyrdquo or conceptstructure for the project incorporates existing taxonomies andthesauri using new standards such as the Resource Descrip-tion Framework and other metadata standards to makeresources sharable and new tools such as Proteacutegeacute for collec-tively developing ontologies

The concept map underlying the Continuum Project is visu-ally represented in several ways A two-dimensional land-scape navigable by the user offers ldquopoints of interestrdquo linked tomultimedia and interactive resources overlaid on geographicmaps of real watersheds Similarly a two-dimensional mapof concepts allows users to explore related ideas and resources

Each of these interfaces to the Continuum content allowsthe user to ldquozoomrdquo into selected subsets of information re-sources which may be perceived as organized collections or asguided pathways (courses) to learning At each point of inter-est the user may zoom in to explore greater detail or zoom outto see a broader view

The ldquozoomingrdquo approach to navigation departs from themore restrictive but generally familiar ldquodesktoprdquo metaphorthat combines pull-down menus buttons and windows In-stead it provides a surrogate for the ldquoreal worldrdquo by structur-ing information spatially showing greater detail as a userzooms ldquocloserrdquo to an object and giving greater perspective (butless detail) as the user zooms farther away In addition thezooming user interface (ZUI) reduces the individual steps auser must take to reach an objective providing a simplerfriendlier navigation system

Using the geographic landscape view a user may ldquozoomrdquoacross a watershed and might take a prearranged tour Start-ing in the headwaters the tour takes a step-by-step walkthrough the channel continuum all the way to the confluenceof the channel with a major waterbody This tour is repeated ata number of different sites to show the diversity of channelprocesses and morphologies in different settings and withdiffering degrees and histories of land management practices

This interactive visual tour of stream channels helps todemonstrate the broad natural variability of stream channelsalong the river continuum (headwaters to mouth of a streamsystem) An improved understanding of the evolution of chan-nel form and processes from the headwaters to the mouth ofthe river will be extremely valuable to managers and regula-tors working on stream buffer-strip designs and riparian man-agement prescriptions

A textual search capability allows a familiar researchmethod for users who have specific objectives and are lessinclined to use the zooming interface

Resources contributed to the Continuum Project are cata-loged using state-of-the-art standards for content-descriptionand content-sharing via the Internet The first version of theContinuum (on CD-ROM) is a visual jumping-off point forlinks to data and resources that adhere to these same openstandards-

The Continuum core team is currently Andy Alm ThomasDunklin Michael Furniss Jeff Guntle Shaun McFarland RileyQuarles Michael Penney Terry Roelofs Bill Trush and JudyWartella Funding is provided by USDA-Forest Service and theCalifornia Department of Forestry and Fire Protection If you wouldlike to participate or learn more contact Michael Furniss at

The ContinuumContinued from Page 33

the little hillside They told their teacher how bareand unhappy it looked The teacher decided to dosomething She ordered little trees from theCalifornia State Forester in Sacramento Shebought some grass seed Each pupil brought tencents to pay for the trees and the seed

They decided to plant the trees in late autumn

Then the winter rains would help them grow Soone day they all went out to the little hillsideThey planted the trees and sowed the grassHow happy this made the little hillside The boysand girls were happy too They had planted aforest It would grow and after many years otherboys and girls would hold picnics there

WMC Networker Summer 2001 39

by Clay Brandowtributariesyahoocom

Tributaries formerly Name Stream and Tributaries isa column dedicated to covering events and accom-plishments in the lives of individuals working in water-shed Keeping up with your fellow watersheders isnot only fun but it is also an important part of net-workingThis time Irsquod like to take a moment to reflect onwatershed networking and the origins of the Water-shed Management Council Email and the Internet aresuch important tools for most us these days itrsquos hardto remember what life was like a couple decades agobefore instant convenient messaging and informationon demand from thousands of sources Back thenisolation both geographic and professional was aproblem for many watershed professionals andadvocatesIn the 1970rsquos many entities andagencies had added a few folks infields related to watershed manage-ment but their own managementwas still dominated by one or twoprofessions A lot of agencies that had added water-shed professionals werenrsquot sure what to do theseodd-ducks That put many of us in the awkwardsometimes frustrating sometimes lonely position beingboth watershed practitioners and sole watershedadvocates In those early years my own way ofdealing with these feelings of isolation and frustrationwas through writing long letters to a few longtimefriends and mentors in longhand no less I told you itwas a long time agoThen came the earliest forms of email My firstexposure to it was in the form of the Forest ServicersquosData General (DG) system about 1985 The fre-quency of my communication with friends and mentorswithin the Forest Service increased dramatically Thishelped more But still my view of the watershedworld was limitedThen shortly thereafter the first conference of whatlater became Watershed Management Council(WMC) was held in November 1986 Volume 1number 1 of the Watershed Management CouncilNewsletter (now known as the WatershedNetworker) was published in November 1987 The

WMC Bylaws were adopted in August 1988 Theadvent of WMC immediately widened my perspectiveexposing me and other members to people we wouldnever have met ideas we would not have encounteredand career opportunities we might not have consideredAnd within a very short time these new people ideasand opportunities were made more accessible via theInternet First through universal email not boundedby agency or entity Second through the World WideWeb

As WMC grows and the membership diversifies thepromise of new people to meet new ideas to considerand new places to practice the art and science ofwatershed management continues and builds At aWMC conference or field meeting you can meet

middot A private consultant from Washington Statemiddot A community organizer form Oregon

middot A watershed professor form Coloradomiddot An extension specialist from

Nevadamiddot A municipal watershed man-

ager from northern Californiamiddot A county watershed planner

from southern Californiamiddot A federal land manager from

New Mexicomiddot A researcher form Idaho

middot A lobbyist from Washington DCmiddot Land owners and managers

middot Advocates and interested citizensmiddot Professionals and technicians

middot And a myriad of other folks from a variety placesall interested in watershed management

Here is the bottom line Even in this highly networkedworld we now live in WMC can help you expand yourpersonal watershed network and help you be moresuccessful and happier at what you do best to promotethe art and science of watershed management If youare not a member yet please join If you are a WMCmember please consider becoming more active in theorganizationAnd remember if youve reached a watershed in yourcareer or have an interesting tidbit of watershed newslet your colleagues know about it Drop a line toTributaries co Clay Brandow 1528 Brown DriveDavis CA 95616 or call me at (916) 653-0719Internet email is best Email me atclay_brandowfirecagov or tributariesyahoocom

Tributaries

Eric Sloane

40 WMC Networker Summer 2001

Thinking Critically at Large Scales 1

Timber Harvest and Sediment Loads in Nine

Northern California Watersheds Based

on Recent TMDL Studies 1

From the President 2

WMC Awards 3

The Watershed Continuum Moving

Towards Consilience 3

Cumulative Watershed EffectsThen and Now 24

Cumulative Watershed Effects Status Gaps

and Needs 34

Insidehellip

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Address Correction Requested

MEMBERSHIP ENROLLMENT

Name _________________________________________

Address _______________________________________

State _________________________________________

Affliation

Phone No

e-mail

q $20 Regular Member q $10 Student

q $50 Institutional q $250 Donorq $30 Outside the USq _____ Years $______year = $______________

Membership dues are for 1 year Non-profit institu-tional members receive 5 copies of each newsletterThe WMC is a non-profit organization All membersreceive subscriptions to WMC publications dis-counts on conference fees and full voting rightsMail this form with your check to

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Cumulative Watershed Effects

Non-Profit Org

US Postage PAID

Davis CA

Permit No 174

ldquoOf all celestial bodies within reach or viewthe most wonderfuland marvelous and mysterious is turning out to be our own planetearth It is the strangest of all places and there is everything in theworld to learn about it It can keep us awake and jubilant withquestions for millennia ahead if we can learn not to meddle and notto destroyrdquo

-Lewis Thomas Late Night Thoughts on Listening toMahlerrsquos Ninth 1984

• stillwaterscicom
• • CDOCUME~1mfurnissDesktopWM

WMC Networker Summer 2001 37

The stems and leaves and roots of the plantsheld the rain The rain stayed on the hillsideinstead of flowing away All summer a littlestream of water trickled down the hillside givinga drink to the thirsty grass and flowers and treeswhen the sun was hot

The birds told the animals about the lovelyhillside and the cool stream and the pleasantshade Many animals came to see--mother deerwith their baby fawns little squirrels and rabbitsand animals that dug homes in the soil andstored their food Fish came to live in the coolstream

The leaves fell andthey made more richsoil Roots held the soiland kept it fromwashing away Thelittle hillside and all ofits friends were happy

Then people found the little valley by thelittle hillside with the flowers and thestream and the fish and the deer Most ofthe people were very thoughtful Theyenjoyed the shade and the flowers and thecool water They ate their lunches on thegreen grass beside the cool stream They

picked up their trash and papersand left the little valley as clean asthey had found it for others toenjoy

But one day somecareless people cameThey made a fire and

had a picnic When they were readyto go home they left their plates andboxes but it didnrsquot really matter verymuch because they left their fireburning too

A little whirlwind came racingdown the hillside He blew the fireover to some leaves--dry leaves-- andthey started to burn The harder thelittle wind blew trying to put the fireout the faster it burned

The next day the little hillside wasvery sad The trees were gone thegrass and the flowers were gone Allof the birds and animals were goneThe little hillside knew the sunwould beat down hot in the summer

One day some boys and girls saw

38 WMC Networker Summer 2001

What problem do we intend to solveThe tendency of the physical and social sciences and the

humanities to break knowledge into its component parts andto focus intently in areas of specialization has led to tremen-dous advances in understanding ndash of parts at the expense ofthe whole The essential problem the Continuum Project ad-dresses is the subject of the ancient Parable of the Blind Menand the Elephanthellip each ldquosawrdquo the elephant based on theirlimited experiences To understand watersheds and to betterunderstand the different perspectives on their managementwe need to connect knowledge across disciplines

Professionals need incentives to continue learning tobroaden their knowledge continually and to provide leader-ship in forming a better collective understanding of water-sheds

Whether it is due to lack of time or competition for atten-tion most people donrsquot read much Most professionals engagedin watershed management canrsquot keep up with the volume ofavailable research and literature A method for rapidly andeffectively disseminating information among professionals isneeded that is engaging enough to attract a larger percentageof the audience than printed materials do now

Society now expects Land management policy decisions tobe effectively informed by applicable scientific findings andthat land managers will take scientific approaches to thecomplex decisions they make and implement To be moreeffective in this experts are needed to filter and synthesize thelarge body of literature available selecting adapting andinterpreting it for land managers

How do we intend to solve itBy mapping key concepts in watershed ecology and linking

them within and across the disciplines they involve the Con-tinuum Project provides a visually exciting multiple-entryresource for professional development It is comfortable forself-paced learners and provides a resource for teachers devel-oping structured outcome-based instruction

Just as modern maps are continually evolving documentswith multiple layers the ldquomappingrdquo of concepts vocabulariesand the relationships between things in the Continuum Projectis open and changeable The overall ldquoontologyrdquo or conceptstructure for the project incorporates existing taxonomies andthesauri using new standards such as the Resource Descrip-tion Framework and other metadata standards to makeresources sharable and new tools such as Proteacutegeacute for collec-tively developing ontologies

The concept map underlying the Continuum Project is visu-ally represented in several ways A two-dimensional land-scape navigable by the user offers ldquopoints of interestrdquo linked tomultimedia and interactive resources overlaid on geographicmaps of real watersheds Similarly a two-dimensional mapof concepts allows users to explore related ideas and resources

Each of these interfaces to the Continuum content allowsthe user to ldquozoomrdquo into selected subsets of information re-sources which may be perceived as organized collections or asguided pathways (courses) to learning At each point of inter-est the user may zoom in to explore greater detail or zoom outto see a broader view

The ldquozoomingrdquo approach to navigation departs from themore restrictive but generally familiar ldquodesktoprdquo metaphorthat combines pull-down menus buttons and windows In-stead it provides a surrogate for the ldquoreal worldrdquo by structur-ing information spatially showing greater detail as a userzooms ldquocloserrdquo to an object and giving greater perspective (butless detail) as the user zooms farther away In addition thezooming user interface (ZUI) reduces the individual steps auser must take to reach an objective providing a simplerfriendlier navigation system

Using the geographic landscape view a user may ldquozoomrdquoacross a watershed and might take a prearranged tour Start-ing in the headwaters the tour takes a step-by-step walkthrough the channel continuum all the way to the confluenceof the channel with a major waterbody This tour is repeated ata number of different sites to show the diversity of channelprocesses and morphologies in different settings and withdiffering degrees and histories of land management practices

This interactive visual tour of stream channels helps todemonstrate the broad natural variability of stream channelsalong the river continuum (headwaters to mouth of a streamsystem) An improved understanding of the evolution of chan-nel form and processes from the headwaters to the mouth ofthe river will be extremely valuable to managers and regula-tors working on stream buffer-strip designs and riparian man-agement prescriptions

A textual search capability allows a familiar researchmethod for users who have specific objectives and are lessinclined to use the zooming interface

Resources contributed to the Continuum Project are cata-loged using state-of-the-art standards for content-descriptionand content-sharing via the Internet The first version of theContinuum (on CD-ROM) is a visual jumping-off point forlinks to data and resources that adhere to these same openstandards-

The Continuum core team is currently Andy Alm ThomasDunklin Michael Furniss Jeff Guntle Shaun McFarland RileyQuarles Michael Penney Terry Roelofs Bill Trush and JudyWartella Funding is provided by USDA-Forest Service and theCalifornia Department of Forestry and Fire Protection If you wouldlike to participate or learn more contact Michael Furniss at

The ContinuumContinued from Page 33

the little hillside They told their teacher how bareand unhappy it looked The teacher decided to dosomething She ordered little trees from theCalifornia State Forester in Sacramento Shebought some grass seed Each pupil brought tencents to pay for the trees and the seed

They decided to plant the trees in late autumn

Then the winter rains would help them grow Soone day they all went out to the little hillsideThey planted the trees and sowed the grassHow happy this made the little hillside The boysand girls were happy too They had planted aforest It would grow and after many years otherboys and girls would hold picnics there

WMC Networker Summer 2001 39

by Clay Brandowtributariesyahoocom

Tributaries formerly Name Stream and Tributaries isa column dedicated to covering events and accom-plishments in the lives of individuals working in water-shed Keeping up with your fellow watersheders isnot only fun but it is also an important part of net-workingThis time Irsquod like to take a moment to reflect onwatershed networking and the origins of the Water-shed Management Council Email and the Internet aresuch important tools for most us these days itrsquos hardto remember what life was like a couple decades agobefore instant convenient messaging and informationon demand from thousands of sources Back thenisolation both geographic and professional was aproblem for many watershed professionals andadvocatesIn the 1970rsquos many entities andagencies had added a few folks infields related to watershed manage-ment but their own managementwas still dominated by one or twoprofessions A lot of agencies that had added water-shed professionals werenrsquot sure what to do theseodd-ducks That put many of us in the awkwardsometimes frustrating sometimes lonely position beingboth watershed practitioners and sole watershedadvocates In those early years my own way ofdealing with these feelings of isolation and frustrationwas through writing long letters to a few longtimefriends and mentors in longhand no less I told you itwas a long time agoThen came the earliest forms of email My firstexposure to it was in the form of the Forest ServicersquosData General (DG) system about 1985 The fre-quency of my communication with friends and mentorswithin the Forest Service increased dramatically Thishelped more But still my view of the watershedworld was limitedThen shortly thereafter the first conference of whatlater became Watershed Management Council(WMC) was held in November 1986 Volume 1number 1 of the Watershed Management CouncilNewsletter (now known as the WatershedNetworker) was published in November 1987 The

WMC Bylaws were adopted in August 1988 Theadvent of WMC immediately widened my perspectiveexposing me and other members to people we wouldnever have met ideas we would not have encounteredand career opportunities we might not have consideredAnd within a very short time these new people ideasand opportunities were made more accessible via theInternet First through universal email not boundedby agency or entity Second through the World WideWeb

As WMC grows and the membership diversifies thepromise of new people to meet new ideas to considerand new places to practice the art and science ofwatershed management continues and builds At aWMC conference or field meeting you can meet

middot A private consultant from Washington Statemiddot A community organizer form Oregon

middot A watershed professor form Coloradomiddot An extension specialist from

Nevadamiddot A municipal watershed man-

ager from northern Californiamiddot A county watershed planner

from southern Californiamiddot A federal land manager from

New Mexicomiddot A researcher form Idaho

middot A lobbyist from Washington DCmiddot Land owners and managers

middot Advocates and interested citizensmiddot Professionals and technicians

middot And a myriad of other folks from a variety placesall interested in watershed management

Here is the bottom line Even in this highly networkedworld we now live in WMC can help you expand yourpersonal watershed network and help you be moresuccessful and happier at what you do best to promotethe art and science of watershed management If youare not a member yet please join If you are a WMCmember please consider becoming more active in theorganizationAnd remember if youve reached a watershed in yourcareer or have an interesting tidbit of watershed newslet your colleagues know about it Drop a line toTributaries co Clay Brandow 1528 Brown DriveDavis CA 95616 or call me at (916) 653-0719Internet email is best Email me atclay_brandowfirecagov or tributariesyahoocom

Tributaries

Eric Sloane

40 WMC Networker Summer 2001

Thinking Critically at Large Scales 1

Timber Harvest and Sediment Loads in Nine

Northern California Watersheds Based

on Recent TMDL Studies 1

From the President 2

WMC Awards 3

The Watershed Continuum Moving

Towards Consilience 3

Cumulative Watershed EffectsThen and Now 24

Cumulative Watershed Effects Status Gaps

and Needs 34

Insidehellip

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Address Correction Requested

MEMBERSHIP ENROLLMENT

Name _________________________________________

Address _______________________________________

State _________________________________________

Affliation

Phone No

e-mail

q $20 Regular Member q $10 Student

q $50 Institutional q $250 Donorq $30 Outside the USq _____ Years $______year = $______________

Membership dues are for 1 year Non-profit institu-tional members receive 5 copies of each newsletterThe WMC is a non-profit organization All membersreceive subscriptions to WMC publications dis-counts on conference fees and full voting rightsMail this form with your check to

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Cumulative Watershed Effects

Non-Profit Org

US Postage PAID

Davis CA

Permit No 174

ldquoOf all celestial bodies within reach or viewthe most wonderfuland marvelous and mysterious is turning out to be our own planetearth It is the strangest of all places and there is everything in theworld to learn about it It can keep us awake and jubilant withquestions for millennia ahead if we can learn not to meddle and notto destroyrdquo

-Lewis Thomas Late Night Thoughts on Listening toMahlerrsquos Ninth 1984

• stillwaterscicom
• • CDOCUME~1mfurnissDesktopWM

38 WMC Networker Summer 2001

What problem do we intend to solveThe tendency of the physical and social sciences and the

humanities to break knowledge into its component parts andto focus intently in areas of specialization has led to tremen-dous advances in understanding ndash of parts at the expense ofthe whole The essential problem the Continuum Project ad-dresses is the subject of the ancient Parable of the Blind Menand the Elephanthellip each ldquosawrdquo the elephant based on theirlimited experiences To understand watersheds and to betterunderstand the different perspectives on their managementwe need to connect knowledge across disciplines

Professionals need incentives to continue learning tobroaden their knowledge continually and to provide leader-ship in forming a better collective understanding of water-sheds

Whether it is due to lack of time or competition for atten-tion most people donrsquot read much Most professionals engagedin watershed management canrsquot keep up with the volume ofavailable research and literature A method for rapidly andeffectively disseminating information among professionals isneeded that is engaging enough to attract a larger percentageof the audience than printed materials do now

Society now expects Land management policy decisions tobe effectively informed by applicable scientific findings andthat land managers will take scientific approaches to thecomplex decisions they make and implement To be moreeffective in this experts are needed to filter and synthesize thelarge body of literature available selecting adapting andinterpreting it for land managers

How do we intend to solve itBy mapping key concepts in watershed ecology and linking

them within and across the disciplines they involve the Con-tinuum Project provides a visually exciting multiple-entryresource for professional development It is comfortable forself-paced learners and provides a resource for teachers devel-oping structured outcome-based instruction

Just as modern maps are continually evolving documentswith multiple layers the ldquomappingrdquo of concepts vocabulariesand the relationships between things in the Continuum Projectis open and changeable The overall ldquoontologyrdquo or conceptstructure for the project incorporates existing taxonomies andthesauri using new standards such as the Resource Descrip-tion Framework and other metadata standards to makeresources sharable and new tools such as Proteacutegeacute for collec-tively developing ontologies

The concept map underlying the Continuum Project is visu-ally represented in several ways A two-dimensional land-scape navigable by the user offers ldquopoints of interestrdquo linked tomultimedia and interactive resources overlaid on geographicmaps of real watersheds Similarly a two-dimensional mapof concepts allows users to explore related ideas and resources

Each of these interfaces to the Continuum content allowsthe user to ldquozoomrdquo into selected subsets of information re-sources which may be perceived as organized collections or asguided pathways (courses) to learning At each point of inter-est the user may zoom in to explore greater detail or zoom outto see a broader view

The ldquozoomingrdquo approach to navigation departs from themore restrictive but generally familiar ldquodesktoprdquo metaphorthat combines pull-down menus buttons and windows In-stead it provides a surrogate for the ldquoreal worldrdquo by structur-ing information spatially showing greater detail as a userzooms ldquocloserrdquo to an object and giving greater perspective (butless detail) as the user zooms farther away In addition thezooming user interface (ZUI) reduces the individual steps auser must take to reach an objective providing a simplerfriendlier navigation system

Using the geographic landscape view a user may ldquozoomrdquoacross a watershed and might take a prearranged tour Start-ing in the headwaters the tour takes a step-by-step walkthrough the channel continuum all the way to the confluenceof the channel with a major waterbody This tour is repeated ata number of different sites to show the diversity of channelprocesses and morphologies in different settings and withdiffering degrees and histories of land management practices

This interactive visual tour of stream channels helps todemonstrate the broad natural variability of stream channelsalong the river continuum (headwaters to mouth of a streamsystem) An improved understanding of the evolution of chan-nel form and processes from the headwaters to the mouth ofthe river will be extremely valuable to managers and regula-tors working on stream buffer-strip designs and riparian man-agement prescriptions

A textual search capability allows a familiar researchmethod for users who have specific objectives and are lessinclined to use the zooming interface

Resources contributed to the Continuum Project are cata-loged using state-of-the-art standards for content-descriptionand content-sharing via the Internet The first version of theContinuum (on CD-ROM) is a visual jumping-off point forlinks to data and resources that adhere to these same openstandards-

The Continuum core team is currently Andy Alm ThomasDunklin Michael Furniss Jeff Guntle Shaun McFarland RileyQuarles Michael Penney Terry Roelofs Bill Trush and JudyWartella Funding is provided by USDA-Forest Service and theCalifornia Department of Forestry and Fire Protection If you wouldlike to participate or learn more contact Michael Furniss at

The ContinuumContinued from Page 33

the little hillside They told their teacher how bareand unhappy it looked The teacher decided to dosomething She ordered little trees from theCalifornia State Forester in Sacramento Shebought some grass seed Each pupil brought tencents to pay for the trees and the seed

They decided to plant the trees in late autumn

Then the winter rains would help them grow Soone day they all went out to the little hillsideThey planted the trees and sowed the grassHow happy this made the little hillside The boysand girls were happy too They had planted aforest It would grow and after many years otherboys and girls would hold picnics there

WMC Networker Summer 2001 39

by Clay Brandowtributariesyahoocom

Tributaries formerly Name Stream and Tributaries isa column dedicated to covering events and accom-plishments in the lives of individuals working in water-shed Keeping up with your fellow watersheders isnot only fun but it is also an important part of net-workingThis time Irsquod like to take a moment to reflect onwatershed networking and the origins of the Water-shed Management Council Email and the Internet aresuch important tools for most us these days itrsquos hardto remember what life was like a couple decades agobefore instant convenient messaging and informationon demand from thousands of sources Back thenisolation both geographic and professional was aproblem for many watershed professionals andadvocatesIn the 1970rsquos many entities andagencies had added a few folks infields related to watershed manage-ment but their own managementwas still dominated by one or twoprofessions A lot of agencies that had added water-shed professionals werenrsquot sure what to do theseodd-ducks That put many of us in the awkwardsometimes frustrating sometimes lonely position beingboth watershed practitioners and sole watershedadvocates In those early years my own way ofdealing with these feelings of isolation and frustrationwas through writing long letters to a few longtimefriends and mentors in longhand no less I told you itwas a long time agoThen came the earliest forms of email My firstexposure to it was in the form of the Forest ServicersquosData General (DG) system about 1985 The fre-quency of my communication with friends and mentorswithin the Forest Service increased dramatically Thishelped more But still my view of the watershedworld was limitedThen shortly thereafter the first conference of whatlater became Watershed Management Council(WMC) was held in November 1986 Volume 1number 1 of the Watershed Management CouncilNewsletter (now known as the WatershedNetworker) was published in November 1987 The

WMC Bylaws were adopted in August 1988 Theadvent of WMC immediately widened my perspectiveexposing me and other members to people we wouldnever have met ideas we would not have encounteredand career opportunities we might not have consideredAnd within a very short time these new people ideasand opportunities were made more accessible via theInternet First through universal email not boundedby agency or entity Second through the World WideWeb

As WMC grows and the membership diversifies thepromise of new people to meet new ideas to considerand new places to practice the art and science ofwatershed management continues and builds At aWMC conference or field meeting you can meet

middot A private consultant from Washington Statemiddot A community organizer form Oregon

middot A watershed professor form Coloradomiddot An extension specialist from

Nevadamiddot A municipal watershed man-

ager from northern Californiamiddot A county watershed planner

from southern Californiamiddot A federal land manager from

New Mexicomiddot A researcher form Idaho

middot A lobbyist from Washington DCmiddot Land owners and managers

middot Advocates and interested citizensmiddot Professionals and technicians

middot And a myriad of other folks from a variety placesall interested in watershed management

Here is the bottom line Even in this highly networkedworld we now live in WMC can help you expand yourpersonal watershed network and help you be moresuccessful and happier at what you do best to promotethe art and science of watershed management If youare not a member yet please join If you are a WMCmember please consider becoming more active in theorganizationAnd remember if youve reached a watershed in yourcareer or have an interesting tidbit of watershed newslet your colleagues know about it Drop a line toTributaries co Clay Brandow 1528 Brown DriveDavis CA 95616 or call me at (916) 653-0719Internet email is best Email me atclay_brandowfirecagov or tributariesyahoocom

Tributaries

Eric Sloane

40 WMC Networker Summer 2001

Thinking Critically at Large Scales 1

Timber Harvest and Sediment Loads in Nine

Northern California Watersheds Based

on Recent TMDL Studies 1

From the President 2

WMC Awards 3

The Watershed Continuum Moving

Towards Consilience 3

Cumulative Watershed EffectsThen and Now 24

Cumulative Watershed Effects Status Gaps

and Needs 34

Insidehellip

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Address Correction Requested

MEMBERSHIP ENROLLMENT

Name _________________________________________

Address _______________________________________

State _________________________________________

Affliation

Phone No

e-mail

q $20 Regular Member q $10 Student

q $50 Institutional q $250 Donorq $30 Outside the USq _____ Years $______year = $______________

Membership dues are for 1 year Non-profit institu-tional members receive 5 copies of each newsletterThe WMC is a non-profit organization All membersreceive subscriptions to WMC publications dis-counts on conference fees and full voting rightsMail this form with your check to

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Cumulative Watershed Effects

Non-Profit Org

US Postage PAID

Davis CA

Permit No 174

ldquoOf all celestial bodies within reach or viewthe most wonderfuland marvelous and mysterious is turning out to be our own planetearth It is the strangest of all places and there is everything in theworld to learn about it It can keep us awake and jubilant withquestions for millennia ahead if we can learn not to meddle and notto destroyrdquo

-Lewis Thomas Late Night Thoughts on Listening toMahlerrsquos Ninth 1984

• stillwaterscicom
• • CDOCUME~1mfurnissDesktopWM

WMC Networker Summer 2001 39

by Clay Brandowtributariesyahoocom

Tributaries formerly Name Stream and Tributaries isa column dedicated to covering events and accom-plishments in the lives of individuals working in water-shed Keeping up with your fellow watersheders isnot only fun but it is also an important part of net-workingThis time Irsquod like to take a moment to reflect onwatershed networking and the origins of the Water-shed Management Council Email and the Internet aresuch important tools for most us these days itrsquos hardto remember what life was like a couple decades agobefore instant convenient messaging and informationon demand from thousands of sources Back thenisolation both geographic and professional was aproblem for many watershed professionals andadvocatesIn the 1970rsquos many entities andagencies had added a few folks infields related to watershed manage-ment but their own managementwas still dominated by one or twoprofessions A lot of agencies that had added water-shed professionals werenrsquot sure what to do theseodd-ducks That put many of us in the awkwardsometimes frustrating sometimes lonely position beingboth watershed practitioners and sole watershedadvocates In those early years my own way ofdealing with these feelings of isolation and frustrationwas through writing long letters to a few longtimefriends and mentors in longhand no less I told you itwas a long time agoThen came the earliest forms of email My firstexposure to it was in the form of the Forest ServicersquosData General (DG) system about 1985 The fre-quency of my communication with friends and mentorswithin the Forest Service increased dramatically Thishelped more But still my view of the watershedworld was limitedThen shortly thereafter the first conference of whatlater became Watershed Management Council(WMC) was held in November 1986 Volume 1number 1 of the Watershed Management CouncilNewsletter (now known as the WatershedNetworker) was published in November 1987 The

WMC Bylaws were adopted in August 1988 Theadvent of WMC immediately widened my perspectiveexposing me and other members to people we wouldnever have met ideas we would not have encounteredand career opportunities we might not have consideredAnd within a very short time these new people ideasand opportunities were made more accessible via theInternet First through universal email not boundedby agency or entity Second through the World WideWeb

As WMC grows and the membership diversifies thepromise of new people to meet new ideas to considerand new places to practice the art and science ofwatershed management continues and builds At aWMC conference or field meeting you can meet

middot A private consultant from Washington Statemiddot A community organizer form Oregon

middot A watershed professor form Coloradomiddot An extension specialist from

Nevadamiddot A municipal watershed man-

ager from northern Californiamiddot A county watershed planner

from southern Californiamiddot A federal land manager from

New Mexicomiddot A researcher form Idaho

middot A lobbyist from Washington DCmiddot Land owners and managers

middot Advocates and interested citizensmiddot Professionals and technicians

middot And a myriad of other folks from a variety placesall interested in watershed management

Here is the bottom line Even in this highly networkedworld we now live in WMC can help you expand yourpersonal watershed network and help you be moresuccessful and happier at what you do best to promotethe art and science of watershed management If youare not a member yet please join If you are a WMCmember please consider becoming more active in theorganizationAnd remember if youve reached a watershed in yourcareer or have an interesting tidbit of watershed newslet your colleagues know about it Drop a line toTributaries co Clay Brandow 1528 Brown DriveDavis CA 95616 or call me at (916) 653-0719Internet email is best Email me atclay_brandowfirecagov or tributariesyahoocom

Tributaries

Eric Sloane

40 WMC Networker Summer 2001

Thinking Critically at Large Scales 1

Timber Harvest and Sediment Loads in Nine

Northern California Watersheds Based

on Recent TMDL Studies 1

From the President 2

WMC Awards 3

The Watershed Continuum Moving

Towards Consilience 3

Cumulative Watershed EffectsThen and Now 24

Cumulative Watershed Effects Status Gaps

and Needs 34

Insidehellip

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Address Correction Requested

MEMBERSHIP ENROLLMENT

Name _________________________________________

Address _______________________________________

State _________________________________________

Affliation

Phone No

e-mail

q $20 Regular Member q $10 Student

q $50 Institutional q $250 Donorq $30 Outside the USq _____ Years $______year = $______________

Membership dues are for 1 year Non-profit institu-tional members receive 5 copies of each newsletterThe WMC is a non-profit organization All membersreceive subscriptions to WMC publications dis-counts on conference fees and full voting rightsMail this form with your check to

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Cumulative Watershed Effects

Non-Profit Org

US Postage PAID

Davis CA

Permit No 174

ldquoOf all celestial bodies within reach or viewthe most wonderfuland marvelous and mysterious is turning out to be our own planetearth It is the strangest of all places and there is everything in theworld to learn about it It can keep us awake and jubilant withquestions for millennia ahead if we can learn not to meddle and notto destroyrdquo

-Lewis Thomas Late Night Thoughts on Listening toMahlerrsquos Ninth 1984

• stillwaterscicom
• • CDOCUME~1mfurnissDesktopWM

40 WMC Networker Summer 2001

Thinking Critically at Large Scales 1

Timber Harvest and Sediment Loads in Nine

Northern California Watersheds Based

on Recent TMDL Studies 1

From the President 2

WMC Awards 3

The Watershed Continuum Moving

Towards Consilience 3

Cumulative Watershed EffectsThen and Now 24

Cumulative Watershed Effects Status Gaps

and Needs 34

Insidehellip

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Address Correction Requested

MEMBERSHIP ENROLLMENT

Name _________________________________________

Address _______________________________________

State _________________________________________

Affliation

Phone No

e-mail

q $20 Regular Member q $10 Student

q $50 Institutional q $250 Donorq $30 Outside the USq _____ Years $______year = $______________

Membership dues are for 1 year Non-profit institu-tional members receive 5 copies of each newsletterThe WMC is a non-profit organization All membersreceive subscriptions to WMC publications dis-counts on conference fees and full voting rightsMail this form with your check to

WATERSHED MANAGEMENT COUNCILco PSRP

University of California at DavisOne Shields Ave

Davis CA 95616

Cumulative Watershed Effects

Non-Profit Org

US Postage PAID

Davis CA

Permit No 174

ldquoOf all celestial bodies within reach or viewthe most wonderfuland marvelous and mysterious is turning out to be our own planetearth It is the strangest of all places and there is everything in theworld to learn about it It can keep us awake and jubilant withquestions for millennia ahead if we can learn not to meddle and notto destroyrdquo

-Lewis Thomas Late Night Thoughts on Listening toMahlerrsquos Ninth 1984

• stillwaterscicom
• • CDOCUME~1mfurnissDesktopWM