In 1935 Arthur Tansley wrote

We cannot confine ourselves to the so-called ldquonaturalrdquo entities

and ignore the processes and expressions of vegetation now so

abundantly provided by man Such a course is not scientifically

sound because scientific analysis must penetrate beneath the

forms of the ldquonaturalrdquo entities and it is not practically useful

because ecology must be applied to conditions brought about by

human activity The ldquonaturalrdquo entities and the anthropogenic

derivates alike must be analyzed in terms of the most appropriate

concepts we can find (Tansley 1935 p 304)

This quote captures the spirit of the new urban emphasisin the US Long-Term Ecological Research (LTER) net-work We know now that Earth abounds with both subtleand pronounced evidence of the influence of people onnatural ecosystems (Russell 1993 Turner and Meyer1993) Arguably cities are the most human dominated ofall ecosystems Recent calls for studies on ldquohuman-domi-nated ecosystemsrdquo (Vitousek et al 1997) finally have beenheeded over 60 years after Tansley penned his warningwith the addition of two metropolises (Phoenix and Balti-more) to the LTER network

In this article we describe an emerging approach tounderstanding the ecology of urban areas by contrastingthese two metropolises and we present a call to action forecologists to integrate their science with that of social sci-entists to achieve a more realistic and useful understand-ing of the natural world in general and its ecology in par-ticular (Pickett and McDonnell 1993 Ehrlich 1997) Webegin by framing a conceptual basis for the study of urbanecological systems the rationale contrasting approachesand special considerations for including human interac-tions at different scales and in a spatial context We thendiscuss the application of our conceptual approach bycomparing site conditions and initial research results inBaltimore and Phoenix We conclude with a summary andsynthesis of implications for the integration of social andecological sciences

The conceptual basis for studying urbanecological systemsWhy has the study of urban ecological systems attracted somuch recent interest The rationale for the study ofhuman-dominated systems is three-pronged Firsthumans dominate Earthrsquos ecosystems (Groffman andLikens 1994 Botsford et al 1997 Chapin et al 1997 Mat-son et al 1997 Noble and Dirzo 1997 Vitousek et al1997) therefore humans must be integrated into modelsfor a complete understanding of extant ecological systemsSecond development of these more realistic models forecological systems will lead to greater success in findingsolutions to environmental problems (Grossmann 1993)Third although the study of ecological phenomena inurban environments is not a new area of science the con-cept of city as ecosystem is relatively new for the field of

July 2000 Vol 50 No 7 BioScience 571

Articles

Nancy B Grimm (e-mail nbgrimmasuedu) is a professor in theDepartment of Biology and Charles L Redman is a professor of

anthropology and director of the Center for Environmental Studies

Arizona State University Tempe AZ 85287 J Morgan Grove is a

research scientist for the USDA Forest Service Northeastern

Research Station South Burlington VT 05403 Steward T A

Pickett is a senior scientist at the Institute of Ecosystem Studies

Millbrook NY 12545

URBAN ECOLOGICAL SYSTEMS PRESENT

MULTIPLE CHALLENGES TO ECOLOGISTSmdash

PERVASIVE HUMAN IMPACT AND EXTREME

HETEROGENEITY OF CITIES AND THE

NEED TO INTEGRATE SOCIAL AND

ECOLOGICAL APPROACHES CONCEPTS

AND THEORY

Integrated Approaches toLong-Term Studies ofUrban Ecological Systems

Integrated Approaches toLong-Term Studies ofUrban Ecological SystemsNANCY B GRIMM J MORGAN GROVE STEWARD T A PICKETT AND CHARLES L REDMAN

ecology Studying cities as ecosystems within new para-digms of ecosystem science (Pickett et al 1992 Wu andLoucks 1995 Flores et al 1997) will both raise the collec-tive consciousness of ecologists regarding urban ecosys-tems and contribute to the further development of con-cepts that apply to all ecosystems

Evidence that human influence on the environment hasbeen pervasive for thousands of years has been accumu-lating from anthropological and archaeological research(Turner et al 1990 Redman 1999) Major human impactson the environment probably go back as far as12000ndash15000 years ago when Siberian hunters firstentered the Americas These hunters may have played animportant role in the extinction of many species of largemammals (Morgan 1993 MacLeish 1994 Stringer andMcKie 1996) The introduction of agriculture trans-formed humanndashenvironmental relations in virtually allparts of the world leading to in addition to the obviousbenefits localized but intense episodes of deforestationsoil erosion disease and regionwide degradation of vege-tative cover long before the modern era (for the Mediter-ranean Butzer 1996 for Central America Rice 1996)Among the more severe human-induced environmentalimpacts are those associated with ancient urban societieswhose dense populations rising rates of consumptionand agricultural intensification led to regional degrada-tion so extreme that cities were abandoned and the pro-ductive potential of entire civilizations was undermined tothe point of ruin Archaeology has documented repeatedexamples of such impacts including the salinization ofsouthern Mesopotamia 4000 years ago (Redman 1992)valleywide erosion in ancient Greece (van Andel et al1990) and almost complete depopulation of large tracts ofGuatemala (Rice 1996) and highland Mexico (OrsquoHara etal 1993) from 1000 AD to 1400 AD Clearly humanactions dramatically alter the functioning of ecosystems ofwhich humans are a part and equally clearly humans area part of virtually all ecosystems and have been so for mil-lennia Nowhere has this human participation been moreintense than in cities suburbs and exurbs and in the sup-porting hinterlands

Today urbanization is a dominant demographic trendand an important component of land-transformationprocesses worldwide Slightly less than half of the worldrsquospopulation now resides in cities but this proportion isprojected to rise to 61 in the next 30 years (UN 1997a)The developed nations have more highly urbanized popu-lations for example close to 80 of the US population isurban However projections for the twenty-first centuryindicate that the largest cities and the largest growth incity size will occur in developing nations Between 1980and 2030 the percentage of the urban population on theAfrican continent will double from 27 to 54 (UN1997a) Urbanization trends of the past century also showa dramatic rise in the size of cities Over 300 cities havemore than 1 million inhabitants and 16 ldquomegacitiesrdquo have

populations exceeding 10 million (UN 1997b) Urbaniza-tion interacts with global change in important ways Forexample although urban areas account for only 2 ofEarthrsquos land surface they produce 78 of greenhouse gas-es thus contributing to global climate change Cities alsoplay a central role in alteration of global biogeochemicalcycles changes in biodiversity due to habitat fragmenta-tion and exotic species and changes in land use and coverfar beyond the cityrsquos boundaries (ie within the urbanldquofootprintrdquo)

The growing impact of urban areas on the face of theearth is reason enough to study them An even more com-pelling argument for understanding how cities work in anecological sense is the fact that humans live in them andmust depend on proper management to maintain anacceptable quality of life for the foreseeable futureBecause human societies are an important part of urbanecological systems (and perhaps all ecosystems McDon-nell and Pickett 1993) ecologists now recognize that ldquomostaspects of the structure and functioning of Earthrsquos ecosys-tems cannot be understood without accounting for thestrong often dominant influence of humanityrdquo (Vitouseket al 1997 p 494)

To understand human actions and influences on ecosys-tems it is essential to use approaches developed in thesocial behavioral and economic sciences Conceptualframeworks that explicitly include humans will be muchmore likely than those that exclude them to accuratelyinform environmental problem solving The reasons forthis contention are obvious Human perception choiceand action are often the phenomena that drive politicaleconomic or cultural decisions that lead to or respond tochange in ecological systems

What can the integration of social and ecological sci-ences as applied to human-dominated ecosystems bring tothe field of ecology One of the main goals of the LTERprogram is to understand the long-term dynamics ofecosystems Although approaches types of ecosystemsand disciplinary expertise often differ among the sitesconceptual similarities in many ways overshadow thesedifferences In their common commitment to understand-ing long-term ecological dynamics for example mostLTERs recognize two classes of variables affecting ecosys-tems The first and better-studied class of variablesincludes patterns and processes of ecosystems which areconstrained by ldquonaturalrdquo factors such as geologic settingclimate and its variation species pools hydrologicprocesses and other biological or geophysical factorsUnderlying this first class of variables are the fundamentaldrivers of ecological systems the flows of energy andinformation and the cycling of matter (by the flow ofinformation we refer primarily to evolutionary originsand change eg Reiners 1986) Understanding how thesedrivers and the constraints they impose interact with eco-logical patterns and processes to produce long-termdynamics has been a major goal of most LTER programs

572 BioScience July 2000 Vol 50 No 7

Articles

The second class of variables arethose directly associated with humanactivities such as land-use changeintroduction or domestication ofspecies consumption of resourcesand production of wastes

The simplified model shown inFigure 1 defines the intellectual arenawithin which LTER ecologists typi-cally work The inclusion of both nat-ural processes and human activitiesinfluencing long-term ecosystemdynamics is appropriate because evenLTER sites in purportedly pristineareas are subject to human-causeddisturbance and change Patterns andprocesses of ecosystems where thefive core areas (primary productionpopulations organic matter nutri-ents and disturbance) of the LTERprogram are centered have beenmore completely specified than havesocial patterns and processes It is clear however that sev-eral important interactions and feedbacks are missingfrom this approach Because many of these missing fea-tures are the subject matter of the social sciences it isthrough contributions in this area that an understandingof the worldrsquos ecosystems can be most enhanced (Ehrlich1997) At the same time these interactions and feedbacksshould not be pursued in isolation within a self-containedtraditional social-scientific framework because thehuman activities that influence ecosystem dynamics arereciprocally influenced both by biogeophysical drivingforces and by ecosystem dynamics

An example of this reciprocal influence between humanactivity and ecosystem dynamics is provided by work atthe North Temperate Lakes LTER site (Carpenter et al1999) one of two LTER programs that have received addi-tional funding to add social science research and regional-ization to their programs (the other augmented programis the Coweeta LTER) In Wisconsin agricultural land usesare linked to eutrophication of the arearsquos lakes primarilythrough excess phosphorus inputs Farmersrsquo use of fertil-izers can directly affect soil dynamics and soil conditionscan in turn affect a farmerrsquos decision to use fertilizers (seeCarpenter et al 1999) In addition a wide set of socioeco-nomic drivers such as the local or regional economy caninfluence human interaction with the natural landscapeIn the example of fertilizer use the market potential of thecrops government subsidies and even the practices ofneighboring farmers can influence a farmerrsquos decision touse fertilizers (see Carpenter et al 1999) Finally ecosys-tem dynamics themselves as altered by impaired waterquality caused by leaching of excess fertilizer-derivednutrients can influence patterns of socioeconomic activi-ty at a larger scale including real estate values industrial

relocation or recreation patterns Examining these inter-actions both complicates and enhances the long-termstudy of an ecosystem

Without understanding interactions and feedbacksbetween human and ecological systems our view ofecosystem dynamics both at local and global scales will belimitedmdashas will be our ability to apply these insights topublic policy and land management Acknowledging thecentral human component leads to an emphasis on newquantitative methods new approaches to modeling newways to account for risk and value the need to understandenvironmental justice and the importance of workingwithin a globally interacting network (eg Grossmann1993) These added interactions and feedbacks tradition-ally have been studied by social scientists in isolation fromlife and earth scientists Rarely if ever has a focused long-term study incorporated all the interactions implicit inFigure 2 Such integration requires a research team thatbrings together scientists from the natural social andengineering sciences in a unified research endeavor

Ecology in and ecology of urban ecological sys-tems The simplest phrase describing the research beingdone in the Baltimore and Phoenix LTER studies is urbanecology However this term is fraught with misunder-standing because it is often applied to research that doesnot encompass the full suite of concepts we hope to devel-op We prefer to distinguish between two different types ofresearch that more accurately describe the projects weenvision ecology in cities and ecology of cities There areabundant examples in the literature of ecology in cities(eg Sukopp and Werner 1982 Sukopp 1990) The basicquestions addressed by such studies are how do ecologicalpatterns and processes differ in cities as compared with

July 2000 Vol 50 No 7 BioScience 573

Articles

Figure 1 Two classes of variables that affect ecosystems patterns and processes ofecosystems and patterns of human activities Ecologists have mostly studied theformer and have developed theory based on the fundamental biogeophysical driversthat determine ecosystem pattern and process Items listed under ldquopatterns andprocesses of ecosystemsrdquo are the core areas of long-term ecological research

other environments What is the effect of the city (ie aconcentration of human population and activities) on theecology of organisms inside and outside of its boundaryand influence (eg McPherson et al 1997) Research top-ics exemplifying ecology in cities are distribution andabundance of animal and plant populations air pollutionand meteorology patch-specific ecological pattern andprocesses edge effects (because edges are especially pro-nounced in cities) and exoticndashnative species interactionsTools for doing ecology in cities include beforendashafterexperiments which allow the study of effects of rapidchanges occurring in the urban environment and the con-cept of urbanndashrural gradients (eg McDonnell et al1993) which can be a form of space-for-time substitutionto detect impacts of urbanization on ecological processesThese examples help illustrate that ecology has been donein cities for a long time We do not wish to claim that theurban LTERs are the inventors of urban ecological studiesBut perhaps much of the uniqueness of the new urbanLTER programs lies in their attempt to treat cities asecosystems that is to study the ecology of cities

The concept of ecology of cities has to do with how theaggregated parts sum that is how cities process energy ormatter relative to their surroundings Ecologists mighttake many different routes to understanding the ecology ofcities mass balances of nutrients for the entire systempatch dynamics (wherein all patch types and the changesin and among them are considered for the landscape as awhole) ecological effects of land-use change whole-sys-tem metabolism spatial distribution of resources and

populations (eg a metapopulationapproach) and estimation of theecological footprint of a city (Reesand Wackernagel 1994 Folke et al1997) Tools for studying ecology ofcities include the watershedapproach wherein measurement ofinputs and outputs is simplifiedbecause the system is defined as thearea of land drained by a particularstream (eg Bormann and Likens1967 Likens and Bormann 1995)patch dynamics modeling andmonitoring and modeling of land-use change by incorporating remotesensing and GIS (geographic infor-mation systems) methodologies Ofcourse each approach borrows fromthe other and with increasing scalewhat is viewed as the ecology-of-cities approach may become ecologyin an urban patch within the largerregion of which the city is part Ourintent is not to advocate one ap-proach over the other but we doagree with McDonnell and Pickett

(1990 p 1231) who stated nearly a decade ago that ldquostudyof the [city] as an ecosystemwould be a radical expan-sion of ecologyrdquo Thus both the Baltimore and PhoenixLTER programs are doing ecology in cities but are framingtheir work within the context of city as ecosystem

An ecosystem is a piece of earth of any size that containsinteracting biotic and abiotic elements and that interactswith its surroundings By this definition and with Tans-leyrsquos purpose and definition of the term in mindmdashldquothewhole system (in the sense of physics) including not onlythe organism-complex but also the whole complex ofphysical factors forming what we call the environment ofthe biomemdashthe habitat factors in the widest senserdquo (Tans-ley 1935 p 299)mdasha city is most certainly an ecosystemBut few studies have treated cities as such (a notableexception is the study of Hong Kong by Boyden et al 1981see also Boyle et al 1997) Ecosystems have definablestructure and function Structure refers to the componentparts of the system organismal (including human) popu-lations landscape patches soils and geologic parent mate-rial and local atmospheric and hydrologic systemsEcosystem function is a general term referring to the suiteof processes such as primary production ecosystem respi-ration biogeochemical transformations informationtransfer and material transport that occur within ecosys-tems and link the structural components The function ofa whole ecosystem or a part of an ecosystem can bethought of as an integrated measure of what that unit doesin the context of its surroundings For example a compo-nent ecosystem in a region contributes stocks of resources

574 BioScience July 2000 Vol 50 No 7

Articles

Figure 2 A more comprehensive view of the drivers interactions and feedbacksaffecting ecosystem dynamics that recognizes in addition to biogeophysicaldrivers socioeconomic drivers that determine patterns of human activity Itemslisted under ldquopatterns of human activitiesrdquo are suggested core areas of long-termsocial science research

or fluxes of materials to that larger region More specifi-cally a city might contribute airborne particulates ornitrogenous or sulfurous gases to ecosystems locateddownwind of it

In addition to the structure function and processestraditionally studied by ecologists in any ecosystem urbansystems also contain the dominant components of socialinstitutions culture and behavior and the built environ-ment An ecology-in-cities study that incorporates thesecomponents might consider for example how irrigationpractices or creation of hydrologic infrastructure (egcanals pipes or storm drains) influences the distributionof insects on household or neighborhood scales An ecol-ogy-of-cities approach might include models of urbangrowth and spread that reflect economic and social dri-vers for example the tendency of people to want to live onhillslopes (behavior) or the market value of housing neartransportation routes (economics) This type of studymust necessarily involve the reconceptualization ofhuman activities not as disturbances to the ecosystem butas important drivers of and limitations to it (eg Padoch1993) Traditional scholarship of urban systems in thehumanndashsocial ecological geophysical and civil infra-structural domains has been pursued in relative isolationwith each developing its own disciplines methodologiesand evaluation tools (Borden 1993) Urban ecosystemstudies can bring elements of these disparate approachestogether underscoring the interdependence of these phe-nomena

Traditional biological or earth sciencendashbased approach-es to ecosystem studies are insufficient for urban systemsbecause of the interaction of social systems with biogeo-physical systems (Borden 1993) Although some haveargued that humans can be treated as just another animalpopulation albeit an important one (but see Padoch1993) the suite of social drivers of urban ecosystemsmdashinformation flow culture and institutionsmdashare not easilymodeled within a traditional population framework(Padoch 1993 Turner and Meyer 1993) Several modifica-tions of this framework are necessary to successfully inte-grate human activity into an ecological model The first isto acknowledge the primary importance of human deci-sion-making in the dynamics of the urban ecosystem Thisdecision-making operates within a broad context of cul-ture information and institutions This modification putsan appropriate emphasis on the differential creation flowand control of information within the human ecosystemCulture the learned patterns of behavior for each particu-lar society or group and institutions the formal structuresthat codify patterns of behavior also are central compo-nents of decision-making and thus are key to understand-ing environmentally relevant decisions

State variables of urban ecosystems must include morethan measures of population size species diversity andenergy flow They must also include measures of state asperceived by humans often referred to as ldquoquality of liferdquo

Educational opportunities cultural resources recreationwealth aesthetics and community health all are factorsthat may differ among cities yet these variables have fewparallels in traditional ecosystem studies

Special considerations for the human compo-nent of urban ecological systems Ecologists mustask whether theories developed for ecological systems inthe presumed absence of human influence will be appro-priate for systems like cities where human dominance isunquestionable We suspect that simple modification ofecological theory will prove unsatisfactory because themodifications we have just discussed deal with aspects ofhuman social systems that are far from simple Althoughincorporation of existing social science models (Pickett etal 1997) into ecological theory provides a starting pointdevelopment of a new integrative ecology that explicitlyincorporates human decisions culture institutions andeconomic systems will ultimately be needed (JamesCollins Ann Kinzig Nancy Grimm William Fagan DianeHope Jianguo Wu and Elizabeth Borer unpublishedmanuscript) At the same time it is incumbent upon socialscientists who hope for a realistic understanding of theurban system to consider biogeophysical feedbacks andinteractions with the ecological system The suite of socialsystem components that we plan to include in our newmodels is listed in the box on page 576 although not all ofthese components may be relevant to any particular mod-el or process

Within the context of the LTER program the subjectmatter for investigation started with a set of five core areasthat are common to all LTER research primary produc-tion populations representing trophic structure organicmatter storage and dynamics nutrient transport anddynamics and disturbance Given as we have suggested(Figure 1) that three fundamental biogeophysical driversof ecosystems are the flow of energy the flow of informa-tion and the cycling of materials what are the comparabledrivers and core areas of social systems Social scientistsworking at the augmented (Coweeta and North TemperateLakes) and urban LTER sites consider individuals to beguided in their activities by the knowledge beliefs valuesand social resources shared with other members of theirsocial system at different levels of organization (eg atfamily community state and national levels) In this per-spective the three fundamental drivers of human ele-ments in the ecosystem (Figure 2) are

Flow of information and knowledge

Incorporation of culturally based values and perceptions

Creation and maintenance of institutions and organi-zations

These drivers condition human activities and decisions(see also Turner and Meyer 1993) Although understand-ing the nature and interaction of these drivers must be the

July 2000 Vol 50 No 7 BioScience 575

Articles

ultimate goal of most inquiries actual investigations maymore often be oriented toward measuring the patterns ofhuman activity these processes create Defining the fol-lowing core topicsmdasheach characterized by its activitiesstructure and historic trajectorymdashis key to a comprehen-sive approach to human ecosystem analysis (Figure 2)

Demographic pattern

Economic system

Power hierarchy

Land use and management

Designed environment

Collectively these core topics serve as guidelines of inquiryanalogous to the five ecological core areas identified earlyin the LTER programrsquos history On the one hand theyreflect the processes operating within the system on theother they are a practical guide for field investigationsThe entire LTER network (not just the urban and aug-mented sites) thus provides an excellent starting point toincorporate social science research that is relevant to andintegrated with studies of ecosystem change

A conceptual scheme for understanding urbanecological systems We have now identified drivers andpatterns of activity in both ecological and social systemsthat must collectively be considered for a full understand-ing of human-dominated ecosystems A more specificconceptual scheme or model for how a study of urban eco-logical systems can be approached is represented by Figure3 This scheme has been modified through recent discus-sions from schemes presented by the Baltimore and

Phoenix LTER groups in grant proposals and various pub-lications (eg Grimm 1997 Pickett et al 1997) indeed itis still evolving The diagram includes a set of variablesthat are linked by interactions and feedbacks

The two variables in the upper cornersmdashcoarse-scaleenvironmental context and societal patterns and process-esmdashcan be viewed as constraints which are the outcomeof operation of the fundamental biogeophysical and soci-etal drivers The coarser-scale environmental contextincludes climate geology history and biogeographicalsetting and societal patterns and processes encompass thesocioeconomic system culture demography and socialinstitutions

The middle two variables are the focus of the new urbanLTER research land use and ecological patterns andprocesses associated with any given land use Land useincorporates both the intent and the reality of how a giv-en parcel within a metropolitan boundary is altered byhuman decisions whereas ecological patterns and process-esmdashenergy flow nutrient cycling the hydrologic cyclespecies distribution abundance and interactions ecosys-tem and landscape structure and disturbancemdashare thosephenomena studied by all LTER teams Land use and asso-ciated ecological conditions are viewed at a single point intime using a hierarchical patch mosaic perspective Land-use change occurs because of development urban renew-al changes in land management or ownership or infra-structure development among other causes

The last two variables in the lower corners of Figure 3mdashchanges in ecological conditions and changes in humanperceptions and attitudesmdashresult from the interactionbetween land-use change and ecological pattern andprocess Changes in ecological conditions represent thenext time step of ecological patterns and processes andchanges in human perceptions and attitudes represent thehuman reaction to either ecological pattern and process orchanges therein as expressed through the filter of humanexperience

The interactions and feedbacks depicted in Figure 3 areintended to reflect temporal dynamics to some extent Asan example land-use decisions are based on the environ-mental and politicondashsocioeconomic context the land useis then perceived as good or bad and this feedback canlead to different human decisions This sequence of deci-sions and changes in land use does not incorporate newecological information (ie the changes in ecological con-dition that result from the land-use change) such a situa-tion is unstable and seems unlikely to lead to a sustainableurban environment A sequence of interactions and feed-backs carried out when a change occurs or in response toan environmental problem however would incorporateeither short-term solutions to those problems or adjust-ments in management decisions based on a solid ecologi-cal foundation

To illustrate the sequence of changes and feedbacks tofurther change consider an example from the Central

576 BioScience July 2000 Vol 50 No 7

Articles

Social institutions Health Justice Faith Commerce Education Leisure Government

Social dynamics Physiological Individual Organizational Institutional Environmental

Social system componentsThe following are examples of social system componentsto be incorporated into humanndashecological models ofurban ecosystems (from Pickett et al 1997)

Social order Age Gender Class Norms Wealth Power Status Knowledge Territory

Social resources Economic (information

population labor) Cultural (organizations

beliefs myths)

ArizonandashPhoenix LTER sitethe establishment of an artifi-cial lake in the once-dry SaltRiver bed (Figure 4) The ini-tial land-use decision (estab-lishment of the lake) was con-strained on the physicalndashecological side by the existenceof an alluvial channel with nosurface water flow (because ofupstream impoundments) in aregion of North America char-acterized by a high propensityfor flash flooding (Baker1977) Societal constraintsincluded the economic cost ofthe project the perceptions ofpolitical and economic bene-fit available technology (col-lapsible dams recirculatingpumps) and the existence ofhuman-created infrastructure(engineered channel diversion ofsurface flow into canals) Giventhese constraints and based onour best understanding of lakeecology the ecological conditionsassociated with the lake when itwas filled were likely to be highnutrients with concomitant highalgal production high rates ofinfiltration and a high probabilityof floods At the next time step weexpected such changes in ecologi-cal conditions as eutrophicationlosses of water to the groundwatersystem and establishment of arobust mosquito population Pre-liminary monitoring confirms thepredictions of high phytoplank-ton biomass and insect popula-tions and a summer flash floodresulted in a brief episode of high phosphorus loading(Amalfi 1999)

The predicted sequence of ecological conditions mayresult in a full range of reactions from the public Feed-backs to the societal patterns and processesmdashin this casethe design and management decisionsmdashcould lead eitherto chemical additions to control algal blooms or to moreecologically based management practices such as controlof nutrient inputs Indeed complaints from the publicabout insect populations already have resulted in imple-mentation of a chemical control program (Amalfi 1999)and feedbacks of ecological information and knowledge ofthe lakersquos dynamics have helped create and modify a lakemanagement plan Thus societal responses can act directly

on the changed conditions (arrow J in Figure 3 eg addi-tion of chemical control agents the technological solutionin Figure 4) or indirectly via an effect on the underlyingecological patterns and processes producing the problem(arrow K in Figure 3 eg diversion of upstream point-source nutrient inputs to the water supply the arrowlabeled management for sustainability in Figure 4) Webelieve that this general scheme (Figure 3) allows us tothink about how social and ecological systems interact inurban areas In addition to addressing fundamental eco-logical processes this approach allows one to make pre-dictions that are testable and it also provides tools that cit-izens and decision-makers can use to create moreecologically sound policy

July 2000 Vol 50 No 7 BioScience 577

Articles

Figure 3 Conceptual scheme for integrating ecological and social systems in urbanenvironments Variables are in boxes interactions and feedbacks are arrows Aenvironmental context sets the range of possibilities for land usendashland cover B societaldecisions and human behavior (incorporating their suite of determinants) are thedirect drivers of land-use change C the pattern of land use (whatever the driver)determines ecological patterns and processes D humans perceive and react to land-usechange (independent of any ecological effects) E humans also perceive and react toecological patterns and processes F in this interaction ecological processes as affectedby land-use change result in a change in ecological conditions G such changes inecological conditions may result in changes in attitudes (even if human perceptionpreviously ignored ecological pattern and process) and changed ecological conditionsare perceived as good or bad by humans H changes in perception and attitude feedback to the societal system (patterns and processes of society) to influence decision-making and this part of the cycle begins anew I in some cases changed ecologicalconditions can alter the coarse-scale environmental context (example urban heatisland) resulting in a feedback that is relatively independent of human response J KWhen a societal response to changed ecological conditions is deemed necessary thesociety can act directly on the changed conditions (J) or on the underlying ecologicalpatterns and processes producing the problem (K) Finally the environmental contextof course influences ecological patterns independent of land use (L)

A modeling framework for investigating citiesThe conceptual insights of the broad scheme in Figure 3illustrate in general how processes from the social realmbiological environment civil infrastructure and the larg-er climatic and geological context can be integrated withthe specific characteristics of metropolitan Phoenix andmetropolitan Baltimore But the extreme patchiness ofurban environments dictates attention to spatial detailusing novel approaches of landscape analysis and model-ing Although the two cities are quite different integra-tion is furthered by a common approach to spatial analy-ses a hierarchical patch dynamics approach (Wu andLoucks 1995 Pickett et al 1999) Hierarchical patchdynamics models start with ecological processes associat-ed with fundamental units of landscapes at some specif-ic scale called patches They then address ecologicalprocesses that are associated with the patches The struc-ture of the patches can be a major determinant of thoseprocesses However patch structure and arrangementalso can change through time Hence models mustaccount for such change that is they must be dynamicFurthermore because patches at a particular scale areoften themselves composed of smaller patches and canbe aggregated into larger patches the models must behierarchical By considering patch dynamics simultane-ously at multiple scales with an accompanying hierarchyof models the complexity of urban systems is rendered

more tractable and translation of information acrossscales is facilitated (Wu 1999)

Special characteristics that dictate a novelframework The modern metropolis presents a striking-ly heterogeneous pattern for study For instance the sharpcontrasts between neighborhoods is a familiar characteris-tic of cities (Clay 1973) Within the span of a city blockwhich is on the order of 200 or fewer meters an observermay cross several obvious boundaries Different kinds ofcommercial use shifts between owner-occupied and rentalproperties and shifts in socioeconomic resources availableto residents are but some of the many contrasts cities pre-sent Such heterogeneity is not unique to dense centralurban districts In fact the contemporary suburb is zonedfor even more discrete transitions than the traditionalmixed use of older cities Residential streets feeder streetscommercial streets strip malls regional malls and indus-trial parks are notable patches in the suburban landscapeOf course the scale of transition in postndashWorld War IIsuburbs tends to be coarser than that of older neighbor-hoods and districts because of the shift to dependence onthe automobile However spatial patchiness in the socialeconomic and infrastructural fabric of metropolitan areasremains their most obvious feature Social scientists havelong been aware of the functional significance of spatial het-erogeneity and mixture of uses within urban areas (Jacobs

578 BioScience July 2000 Vol 50 No 7

Articles

Figure 4 Conceptual scheme adapted to show the interaction among physical ecological engineering social andmanagement variables and drivers in the new Tempe Town Lake Arizona

1962) The ecological significance of such socioeconomicheterogeneity is however an open question Indeed deter-mining to what extent the well-recognized patchiness ofurban areas has ecological dimensions and ecologicalimplications is one of the main motivations of integratedlong-term ecological research in the metropolis

Whatever its ecological significance the conspicuousspatial heterogeneity in urban systems is an entry point forintegration with social science The existence of suchclearly defined patches as neighborhoods and cityscapeswhich combine infrastructural and natural features isapparent to all researchers who must work together togenerate the interdisciplinary synthesis for understandingcities as ecological systems Hydrologists ecologistsdemographers economists engineers and citizens all canand do recognize the spatial heterogeneity of cities Neigh-borhood associations watershed associations censustracts and similar groupings are institutional expressionsof this common recognition Of course each discipline orconstituency may see the boundaries or the most salientfeatures of the patchwork somewhat differently For exam-ple the civil engineer and the urban recreationist will havedifferent views of the boundaries of a watershed The firstmay see a ldquosewershedrdquo the second a visually unified land-scape that is engaging on a morningjog Therefore new multidimensionalclassifications of the heterogeneity ofthe metropolis are required Amongthe principal dimensions of such classi-fications however will be factors thatcontrol the flow of materials energyand information through and withinthe metropolis An emphasis on thesekinds of variables suggests that a spatially explicit ecosys-tem perspective can emerge from the heterogeneity of themetropolis

The interdisciplinary integration required to under-stand the ecological significance of spatial pattern inurban ecosystems must account for several important fea-tures of humans and their institutions Although these fea-tures are implied in the social drivers and phenomena weintroduced earlier (Figure 2 see box page 576) these fea-tures add complexity to ecological studies of the city andtherefore cannot be ignored The spatial heterogeneity ofurban systems is established by formal institutions such aszoning regulations and maintained by other formal insti-tutions such as public works and the courts However lessformal institutions such as families and community asso-ciations also contribute to the spatial structure and itsfunction Humans as individuals and groups are self-aware capable of learning quickly and engaged in exten-sive networks of rapid communication These features ofthe human components of urban systems mean that thefeedback among the biological human infrastructuraland the larger physical contexts can be strong and inmany cases rapid This is one reason that education has

been incorporated into the structure of urban LTER pro-grams We hypothesize that learning about the hetero-geneity and function of an urban area can be a tool thatcitizens and institutions can demonstrably use to improvetheir neighborhoods city and region through manage-ment planning and policy (Grove and Burch 1997)

Patch dynamics and hierarchical approachesSpatial heterogeneity was independently chosen as a start-ing point by both the Central ArizonandashPhoenix and Balti-more Ecosystem Study LTER teams In addition to theadvantages already laid out the patch dynamics approachalso brings the advantage of hierarchical nesting andaggregation Such a flexible hierarchical approach isimportant because the structures and consequently theprocesses that govern the function of the metropolis as anecological system occur at a variety of scales For examplejust because some social processes occur at the scale ofneighborhoods of say 15 square blocks does not meanthat the most important ecological impacts occur at thesame scale Similarly the concentration of resources orwastes in particular patches can be due to decisions madeat great distances from the point of concentration There-fore patches may be scaled up or down for different func-

tional analyses and the configurationof patches that are relevant to specificpaths along which resources and infor-mation flow can be assessed

As an example Baltimore can bedescribed in terms of the five-countymetropolitan area (Figure 5) Within itare three principal watersheds thatextend from the rural hinterlands to the

central city Each of these watersheds can be divided intosubcatchments The 17000-ha Gwynns Falls catchmentcontains 16 smaller watersheds that have been used fordiscussion of management and restoration activities Stillsmaller units can be used for mechanistic studies ofecosystem and socioeconomic processes The fact that dif-ferent nestings are possible within the larger units meansthat the scales at which important interactions occur canbe captured and the promulgation of their effects to dif-ferent scales whether coarser or finer can be determined(see Grove and Burch 1997)

A similar nested set of units is being identified in theCentral ArizonandashPhoenix study area (Figure 5) At thelargest scale three patches exist along an eastndashwest gradi-ent from primarily commercialndashindustrialndashresidential toprimarily agricultural to primarily desert land usendashlandcover The political boundaries of the 24 different munic-ipalities in the Phoenix metropolitan area form anotherset of smaller patches and heterogeneity of land usendashlandcover is evident within these smaller units Interestinglypatch size regularity and connectivity differ within thethree broadest patches (Matthew Luck and Jianguo WuArizona State University personal communication)

July 2000 Vol 50 No 7 BioScience 579

Articles

ldquoThe civil engineer and the

urban recreationist will have

different views of the

boundaries of a watershedrdquo

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

ecology Studying cities as ecosystems within new para-digms of ecosystem science (Pickett et al 1992 Wu andLoucks 1995 Flores et al 1997) will both raise the collec-tive consciousness of ecologists regarding urban ecosys-tems and contribute to the further development of con-cepts that apply to all ecosystems

Evidence that human influence on the environment hasbeen pervasive for thousands of years has been accumu-lating from anthropological and archaeological research(Turner et al 1990 Redman 1999) Major human impactson the environment probably go back as far as12000ndash15000 years ago when Siberian hunters firstentered the Americas These hunters may have played animportant role in the extinction of many species of largemammals (Morgan 1993 MacLeish 1994 Stringer andMcKie 1996) The introduction of agriculture trans-formed humanndashenvironmental relations in virtually allparts of the world leading to in addition to the obviousbenefits localized but intense episodes of deforestationsoil erosion disease and regionwide degradation of vege-tative cover long before the modern era (for the Mediter-ranean Butzer 1996 for Central America Rice 1996)Among the more severe human-induced environmentalimpacts are those associated with ancient urban societieswhose dense populations rising rates of consumptionand agricultural intensification led to regional degrada-tion so extreme that cities were abandoned and the pro-ductive potential of entire civilizations was undermined tothe point of ruin Archaeology has documented repeatedexamples of such impacts including the salinization ofsouthern Mesopotamia 4000 years ago (Redman 1992)valleywide erosion in ancient Greece (van Andel et al1990) and almost complete depopulation of large tracts ofGuatemala (Rice 1996) and highland Mexico (OrsquoHara etal 1993) from 1000 AD to 1400 AD Clearly humanactions dramatically alter the functioning of ecosystems ofwhich humans are a part and equally clearly humans area part of virtually all ecosystems and have been so for mil-lennia Nowhere has this human participation been moreintense than in cities suburbs and exurbs and in the sup-porting hinterlands

Today urbanization is a dominant demographic trendand an important component of land-transformationprocesses worldwide Slightly less than half of the worldrsquospopulation now resides in cities but this proportion isprojected to rise to 61 in the next 30 years (UN 1997a)The developed nations have more highly urbanized popu-lations for example close to 80 of the US population isurban However projections for the twenty-first centuryindicate that the largest cities and the largest growth incity size will occur in developing nations Between 1980and 2030 the percentage of the urban population on theAfrican continent will double from 27 to 54 (UN1997a) Urbanization trends of the past century also showa dramatic rise in the size of cities Over 300 cities havemore than 1 million inhabitants and 16 ldquomegacitiesrdquo have

populations exceeding 10 million (UN 1997b) Urbaniza-tion interacts with global change in important ways Forexample although urban areas account for only 2 ofEarthrsquos land surface they produce 78 of greenhouse gas-es thus contributing to global climate change Cities alsoplay a central role in alteration of global biogeochemicalcycles changes in biodiversity due to habitat fragmenta-tion and exotic species and changes in land use and coverfar beyond the cityrsquos boundaries (ie within the urbanldquofootprintrdquo)

The growing impact of urban areas on the face of theearth is reason enough to study them An even more com-pelling argument for understanding how cities work in anecological sense is the fact that humans live in them andmust depend on proper management to maintain anacceptable quality of life for the foreseeable futureBecause human societies are an important part of urbanecological systems (and perhaps all ecosystems McDon-nell and Pickett 1993) ecologists now recognize that ldquomostaspects of the structure and functioning of Earthrsquos ecosys-tems cannot be understood without accounting for thestrong often dominant influence of humanityrdquo (Vitouseket al 1997 p 494)

To understand human actions and influences on ecosys-tems it is essential to use approaches developed in thesocial behavioral and economic sciences Conceptualframeworks that explicitly include humans will be muchmore likely than those that exclude them to accuratelyinform environmental problem solving The reasons forthis contention are obvious Human perception choiceand action are often the phenomena that drive politicaleconomic or cultural decisions that lead to or respond tochange in ecological systems

What can the integration of social and ecological sci-ences as applied to human-dominated ecosystems bring tothe field of ecology One of the main goals of the LTERprogram is to understand the long-term dynamics ofecosystems Although approaches types of ecosystemsand disciplinary expertise often differ among the sitesconceptual similarities in many ways overshadow thesedifferences In their common commitment to understand-ing long-term ecological dynamics for example mostLTERs recognize two classes of variables affecting ecosys-tems The first and better-studied class of variablesincludes patterns and processes of ecosystems which areconstrained by ldquonaturalrdquo factors such as geologic settingclimate and its variation species pools hydrologicprocesses and other biological or geophysical factorsUnderlying this first class of variables are the fundamentaldrivers of ecological systems the flows of energy andinformation and the cycling of matter (by the flow ofinformation we refer primarily to evolutionary originsand change eg Reiners 1986) Understanding how thesedrivers and the constraints they impose interact with eco-logical patterns and processes to produce long-termdynamics has been a major goal of most LTER programs

572 BioScience July 2000 Vol 50 No 7

Articles

The second class of variables arethose directly associated with humanactivities such as land-use changeintroduction or domestication ofspecies consumption of resourcesand production of wastes

The simplified model shown inFigure 1 defines the intellectual arenawithin which LTER ecologists typi-cally work The inclusion of both nat-ural processes and human activitiesinfluencing long-term ecosystemdynamics is appropriate because evenLTER sites in purportedly pristineareas are subject to human-causeddisturbance and change Patterns andprocesses of ecosystems where thefive core areas (primary productionpopulations organic matter nutri-ents and disturbance) of the LTERprogram are centered have beenmore completely specified than havesocial patterns and processes It is clear however that sev-eral important interactions and feedbacks are missingfrom this approach Because many of these missing fea-tures are the subject matter of the social sciences it isthrough contributions in this area that an understandingof the worldrsquos ecosystems can be most enhanced (Ehrlich1997) At the same time these interactions and feedbacksshould not be pursued in isolation within a self-containedtraditional social-scientific framework because thehuman activities that influence ecosystem dynamics arereciprocally influenced both by biogeophysical drivingforces and by ecosystem dynamics

An example of this reciprocal influence between humanactivity and ecosystem dynamics is provided by work atthe North Temperate Lakes LTER site (Carpenter et al1999) one of two LTER programs that have received addi-tional funding to add social science research and regional-ization to their programs (the other augmented programis the Coweeta LTER) In Wisconsin agricultural land usesare linked to eutrophication of the arearsquos lakes primarilythrough excess phosphorus inputs Farmersrsquo use of fertil-izers can directly affect soil dynamics and soil conditionscan in turn affect a farmerrsquos decision to use fertilizers (seeCarpenter et al 1999) In addition a wide set of socioeco-nomic drivers such as the local or regional economy caninfluence human interaction with the natural landscapeIn the example of fertilizer use the market potential of thecrops government subsidies and even the practices ofneighboring farmers can influence a farmerrsquos decision touse fertilizers (see Carpenter et al 1999) Finally ecosys-tem dynamics themselves as altered by impaired waterquality caused by leaching of excess fertilizer-derivednutrients can influence patterns of socioeconomic activi-ty at a larger scale including real estate values industrial

relocation or recreation patterns Examining these inter-actions both complicates and enhances the long-termstudy of an ecosystem

Without understanding interactions and feedbacksbetween human and ecological systems our view ofecosystem dynamics both at local and global scales will belimitedmdashas will be our ability to apply these insights topublic policy and land management Acknowledging thecentral human component leads to an emphasis on newquantitative methods new approaches to modeling newways to account for risk and value the need to understandenvironmental justice and the importance of workingwithin a globally interacting network (eg Grossmann1993) These added interactions and feedbacks tradition-ally have been studied by social scientists in isolation fromlife and earth scientists Rarely if ever has a focused long-term study incorporated all the interactions implicit inFigure 2 Such integration requires a research team thatbrings together scientists from the natural social andengineering sciences in a unified research endeavor

Ecology in and ecology of urban ecological sys-tems The simplest phrase describing the research beingdone in the Baltimore and Phoenix LTER studies is urbanecology However this term is fraught with misunder-standing because it is often applied to research that doesnot encompass the full suite of concepts we hope to devel-op We prefer to distinguish between two different types ofresearch that more accurately describe the projects weenvision ecology in cities and ecology of cities There areabundant examples in the literature of ecology in cities(eg Sukopp and Werner 1982 Sukopp 1990) The basicquestions addressed by such studies are how do ecologicalpatterns and processes differ in cities as compared with

July 2000 Vol 50 No 7 BioScience 573

Articles

Figure 1 Two classes of variables that affect ecosystems patterns and processes ofecosystems and patterns of human activities Ecologists have mostly studied theformer and have developed theory based on the fundamental biogeophysical driversthat determine ecosystem pattern and process Items listed under ldquopatterns andprocesses of ecosystemsrdquo are the core areas of long-term ecological research

other environments What is the effect of the city (ie aconcentration of human population and activities) on theecology of organisms inside and outside of its boundaryand influence (eg McPherson et al 1997) Research top-ics exemplifying ecology in cities are distribution andabundance of animal and plant populations air pollutionand meteorology patch-specific ecological pattern andprocesses edge effects (because edges are especially pro-nounced in cities) and exoticndashnative species interactionsTools for doing ecology in cities include beforendashafterexperiments which allow the study of effects of rapidchanges occurring in the urban environment and the con-cept of urbanndashrural gradients (eg McDonnell et al1993) which can be a form of space-for-time substitutionto detect impacts of urbanization on ecological processesThese examples help illustrate that ecology has been donein cities for a long time We do not wish to claim that theurban LTERs are the inventors of urban ecological studiesBut perhaps much of the uniqueness of the new urbanLTER programs lies in their attempt to treat cities asecosystems that is to study the ecology of cities

The concept of ecology of cities has to do with how theaggregated parts sum that is how cities process energy ormatter relative to their surroundings Ecologists mighttake many different routes to understanding the ecology ofcities mass balances of nutrients for the entire systempatch dynamics (wherein all patch types and the changesin and among them are considered for the landscape as awhole) ecological effects of land-use change whole-sys-tem metabolism spatial distribution of resources and

populations (eg a metapopulationapproach) and estimation of theecological footprint of a city (Reesand Wackernagel 1994 Folke et al1997) Tools for studying ecology ofcities include the watershedapproach wherein measurement ofinputs and outputs is simplifiedbecause the system is defined as thearea of land drained by a particularstream (eg Bormann and Likens1967 Likens and Bormann 1995)patch dynamics modeling andmonitoring and modeling of land-use change by incorporating remotesensing and GIS (geographic infor-mation systems) methodologies Ofcourse each approach borrows fromthe other and with increasing scalewhat is viewed as the ecology-of-cities approach may become ecologyin an urban patch within the largerregion of which the city is part Ourintent is not to advocate one ap-proach over the other but we doagree with McDonnell and Pickett

(1990 p 1231) who stated nearly a decade ago that ldquostudyof the [city] as an ecosystemwould be a radical expan-sion of ecologyrdquo Thus both the Baltimore and PhoenixLTER programs are doing ecology in cities but are framingtheir work within the context of city as ecosystem

An ecosystem is a piece of earth of any size that containsinteracting biotic and abiotic elements and that interactswith its surroundings By this definition and with Tans-leyrsquos purpose and definition of the term in mindmdashldquothewhole system (in the sense of physics) including not onlythe organism-complex but also the whole complex ofphysical factors forming what we call the environment ofthe biomemdashthe habitat factors in the widest senserdquo (Tans-ley 1935 p 299)mdasha city is most certainly an ecosystemBut few studies have treated cities as such (a notableexception is the study of Hong Kong by Boyden et al 1981see also Boyle et al 1997) Ecosystems have definablestructure and function Structure refers to the componentparts of the system organismal (including human) popu-lations landscape patches soils and geologic parent mate-rial and local atmospheric and hydrologic systemsEcosystem function is a general term referring to the suiteof processes such as primary production ecosystem respi-ration biogeochemical transformations informationtransfer and material transport that occur within ecosys-tems and link the structural components The function ofa whole ecosystem or a part of an ecosystem can bethought of as an integrated measure of what that unit doesin the context of its surroundings For example a compo-nent ecosystem in a region contributes stocks of resources

574 BioScience July 2000 Vol 50 No 7

Articles

Figure 2 A more comprehensive view of the drivers interactions and feedbacksaffecting ecosystem dynamics that recognizes in addition to biogeophysicaldrivers socioeconomic drivers that determine patterns of human activity Itemslisted under ldquopatterns of human activitiesrdquo are suggested core areas of long-termsocial science research

or fluxes of materials to that larger region More specifi-cally a city might contribute airborne particulates ornitrogenous or sulfurous gases to ecosystems locateddownwind of it

In addition to the structure function and processestraditionally studied by ecologists in any ecosystem urbansystems also contain the dominant components of socialinstitutions culture and behavior and the built environ-ment An ecology-in-cities study that incorporates thesecomponents might consider for example how irrigationpractices or creation of hydrologic infrastructure (egcanals pipes or storm drains) influences the distributionof insects on household or neighborhood scales An ecol-ogy-of-cities approach might include models of urbangrowth and spread that reflect economic and social dri-vers for example the tendency of people to want to live onhillslopes (behavior) or the market value of housing neartransportation routes (economics) This type of studymust necessarily involve the reconceptualization ofhuman activities not as disturbances to the ecosystem butas important drivers of and limitations to it (eg Padoch1993) Traditional scholarship of urban systems in thehumanndashsocial ecological geophysical and civil infra-structural domains has been pursued in relative isolationwith each developing its own disciplines methodologiesand evaluation tools (Borden 1993) Urban ecosystemstudies can bring elements of these disparate approachestogether underscoring the interdependence of these phe-nomena

Traditional biological or earth sciencendashbased approach-es to ecosystem studies are insufficient for urban systemsbecause of the interaction of social systems with biogeo-physical systems (Borden 1993) Although some haveargued that humans can be treated as just another animalpopulation albeit an important one (but see Padoch1993) the suite of social drivers of urban ecosystemsmdashinformation flow culture and institutionsmdashare not easilymodeled within a traditional population framework(Padoch 1993 Turner and Meyer 1993) Several modifica-tions of this framework are necessary to successfully inte-grate human activity into an ecological model The first isto acknowledge the primary importance of human deci-sion-making in the dynamics of the urban ecosystem Thisdecision-making operates within a broad context of cul-ture information and institutions This modification putsan appropriate emphasis on the differential creation flowand control of information within the human ecosystemCulture the learned patterns of behavior for each particu-lar society or group and institutions the formal structuresthat codify patterns of behavior also are central compo-nents of decision-making and thus are key to understand-ing environmentally relevant decisions

State variables of urban ecosystems must include morethan measures of population size species diversity andenergy flow They must also include measures of state asperceived by humans often referred to as ldquoquality of liferdquo

Educational opportunities cultural resources recreationwealth aesthetics and community health all are factorsthat may differ among cities yet these variables have fewparallels in traditional ecosystem studies

Special considerations for the human compo-nent of urban ecological systems Ecologists mustask whether theories developed for ecological systems inthe presumed absence of human influence will be appro-priate for systems like cities where human dominance isunquestionable We suspect that simple modification ofecological theory will prove unsatisfactory because themodifications we have just discussed deal with aspects ofhuman social systems that are far from simple Althoughincorporation of existing social science models (Pickett etal 1997) into ecological theory provides a starting pointdevelopment of a new integrative ecology that explicitlyincorporates human decisions culture institutions andeconomic systems will ultimately be needed (JamesCollins Ann Kinzig Nancy Grimm William Fagan DianeHope Jianguo Wu and Elizabeth Borer unpublishedmanuscript) At the same time it is incumbent upon socialscientists who hope for a realistic understanding of theurban system to consider biogeophysical feedbacks andinteractions with the ecological system The suite of socialsystem components that we plan to include in our newmodels is listed in the box on page 576 although not all ofthese components may be relevant to any particular mod-el or process

Within the context of the LTER program the subjectmatter for investigation started with a set of five core areasthat are common to all LTER research primary produc-tion populations representing trophic structure organicmatter storage and dynamics nutrient transport anddynamics and disturbance Given as we have suggested(Figure 1) that three fundamental biogeophysical driversof ecosystems are the flow of energy the flow of informa-tion and the cycling of materials what are the comparabledrivers and core areas of social systems Social scientistsworking at the augmented (Coweeta and North TemperateLakes) and urban LTER sites consider individuals to beguided in their activities by the knowledge beliefs valuesand social resources shared with other members of theirsocial system at different levels of organization (eg atfamily community state and national levels) In this per-spective the three fundamental drivers of human ele-ments in the ecosystem (Figure 2) are

Flow of information and knowledge

Incorporation of culturally based values and perceptions

Creation and maintenance of institutions and organi-zations

These drivers condition human activities and decisions(see also Turner and Meyer 1993) Although understand-ing the nature and interaction of these drivers must be the

July 2000 Vol 50 No 7 BioScience 575

Articles

ultimate goal of most inquiries actual investigations maymore often be oriented toward measuring the patterns ofhuman activity these processes create Defining the fol-lowing core topicsmdasheach characterized by its activitiesstructure and historic trajectorymdashis key to a comprehen-sive approach to human ecosystem analysis (Figure 2)

Demographic pattern

Economic system

Power hierarchy

Land use and management

Designed environment

Collectively these core topics serve as guidelines of inquiryanalogous to the five ecological core areas identified earlyin the LTER programrsquos history On the one hand theyreflect the processes operating within the system on theother they are a practical guide for field investigationsThe entire LTER network (not just the urban and aug-mented sites) thus provides an excellent starting point toincorporate social science research that is relevant to andintegrated with studies of ecosystem change

A conceptual scheme for understanding urbanecological systems We have now identified drivers andpatterns of activity in both ecological and social systemsthat must collectively be considered for a full understand-ing of human-dominated ecosystems A more specificconceptual scheme or model for how a study of urban eco-logical systems can be approached is represented by Figure3 This scheme has been modified through recent discus-sions from schemes presented by the Baltimore and

Phoenix LTER groups in grant proposals and various pub-lications (eg Grimm 1997 Pickett et al 1997) indeed itis still evolving The diagram includes a set of variablesthat are linked by interactions and feedbacks

The two variables in the upper cornersmdashcoarse-scaleenvironmental context and societal patterns and process-esmdashcan be viewed as constraints which are the outcomeof operation of the fundamental biogeophysical and soci-etal drivers The coarser-scale environmental contextincludes climate geology history and biogeographicalsetting and societal patterns and processes encompass thesocioeconomic system culture demography and socialinstitutions

The middle two variables are the focus of the new urbanLTER research land use and ecological patterns andprocesses associated with any given land use Land useincorporates both the intent and the reality of how a giv-en parcel within a metropolitan boundary is altered byhuman decisions whereas ecological patterns and process-esmdashenergy flow nutrient cycling the hydrologic cyclespecies distribution abundance and interactions ecosys-tem and landscape structure and disturbancemdashare thosephenomena studied by all LTER teams Land use and asso-ciated ecological conditions are viewed at a single point intime using a hierarchical patch mosaic perspective Land-use change occurs because of development urban renew-al changes in land management or ownership or infra-structure development among other causes

The last two variables in the lower corners of Figure 3mdashchanges in ecological conditions and changes in humanperceptions and attitudesmdashresult from the interactionbetween land-use change and ecological pattern andprocess Changes in ecological conditions represent thenext time step of ecological patterns and processes andchanges in human perceptions and attitudes represent thehuman reaction to either ecological pattern and process orchanges therein as expressed through the filter of humanexperience

The interactions and feedbacks depicted in Figure 3 areintended to reflect temporal dynamics to some extent Asan example land-use decisions are based on the environ-mental and politicondashsocioeconomic context the land useis then perceived as good or bad and this feedback canlead to different human decisions This sequence of deci-sions and changes in land use does not incorporate newecological information (ie the changes in ecological con-dition that result from the land-use change) such a situa-tion is unstable and seems unlikely to lead to a sustainableurban environment A sequence of interactions and feed-backs carried out when a change occurs or in response toan environmental problem however would incorporateeither short-term solutions to those problems or adjust-ments in management decisions based on a solid ecologi-cal foundation

To illustrate the sequence of changes and feedbacks tofurther change consider an example from the Central

576 BioScience July 2000 Vol 50 No 7

Articles

Social institutions Health Justice Faith Commerce Education Leisure Government

Social dynamics Physiological Individual Organizational Institutional Environmental

Social system componentsThe following are examples of social system componentsto be incorporated into humanndashecological models ofurban ecosystems (from Pickett et al 1997)

Social order Age Gender Class Norms Wealth Power Status Knowledge Territory

Social resources Economic (information

population labor) Cultural (organizations

beliefs myths)

ArizonandashPhoenix LTER sitethe establishment of an artifi-cial lake in the once-dry SaltRiver bed (Figure 4) The ini-tial land-use decision (estab-lishment of the lake) was con-strained on the physicalndashecological side by the existenceof an alluvial channel with nosurface water flow (because ofupstream impoundments) in aregion of North America char-acterized by a high propensityfor flash flooding (Baker1977) Societal constraintsincluded the economic cost ofthe project the perceptions ofpolitical and economic bene-fit available technology (col-lapsible dams recirculatingpumps) and the existence ofhuman-created infrastructure(engineered channel diversion ofsurface flow into canals) Giventhese constraints and based onour best understanding of lakeecology the ecological conditionsassociated with the lake when itwas filled were likely to be highnutrients with concomitant highalgal production high rates ofinfiltration and a high probabilityof floods At the next time step weexpected such changes in ecologi-cal conditions as eutrophicationlosses of water to the groundwatersystem and establishment of arobust mosquito population Pre-liminary monitoring confirms thepredictions of high phytoplank-ton biomass and insect popula-tions and a summer flash floodresulted in a brief episode of high phosphorus loading(Amalfi 1999)

The predicted sequence of ecological conditions mayresult in a full range of reactions from the public Feed-backs to the societal patterns and processesmdashin this casethe design and management decisionsmdashcould lead eitherto chemical additions to control algal blooms or to moreecologically based management practices such as controlof nutrient inputs Indeed complaints from the publicabout insect populations already have resulted in imple-mentation of a chemical control program (Amalfi 1999)and feedbacks of ecological information and knowledge ofthe lakersquos dynamics have helped create and modify a lakemanagement plan Thus societal responses can act directly

on the changed conditions (arrow J in Figure 3 eg addi-tion of chemical control agents the technological solutionin Figure 4) or indirectly via an effect on the underlyingecological patterns and processes producing the problem(arrow K in Figure 3 eg diversion of upstream point-source nutrient inputs to the water supply the arrowlabeled management for sustainability in Figure 4) Webelieve that this general scheme (Figure 3) allows us tothink about how social and ecological systems interact inurban areas In addition to addressing fundamental eco-logical processes this approach allows one to make pre-dictions that are testable and it also provides tools that cit-izens and decision-makers can use to create moreecologically sound policy

July 2000 Vol 50 No 7 BioScience 577

Articles

Figure 3 Conceptual scheme for integrating ecological and social systems in urbanenvironments Variables are in boxes interactions and feedbacks are arrows Aenvironmental context sets the range of possibilities for land usendashland cover B societaldecisions and human behavior (incorporating their suite of determinants) are thedirect drivers of land-use change C the pattern of land use (whatever the driver)determines ecological patterns and processes D humans perceive and react to land-usechange (independent of any ecological effects) E humans also perceive and react toecological patterns and processes F in this interaction ecological processes as affectedby land-use change result in a change in ecological conditions G such changes inecological conditions may result in changes in attitudes (even if human perceptionpreviously ignored ecological pattern and process) and changed ecological conditionsare perceived as good or bad by humans H changes in perception and attitude feedback to the societal system (patterns and processes of society) to influence decision-making and this part of the cycle begins anew I in some cases changed ecologicalconditions can alter the coarse-scale environmental context (example urban heatisland) resulting in a feedback that is relatively independent of human response J KWhen a societal response to changed ecological conditions is deemed necessary thesociety can act directly on the changed conditions (J) or on the underlying ecologicalpatterns and processes producing the problem (K) Finally the environmental contextof course influences ecological patterns independent of land use (L)

A modeling framework for investigating citiesThe conceptual insights of the broad scheme in Figure 3illustrate in general how processes from the social realmbiological environment civil infrastructure and the larg-er climatic and geological context can be integrated withthe specific characteristics of metropolitan Phoenix andmetropolitan Baltimore But the extreme patchiness ofurban environments dictates attention to spatial detailusing novel approaches of landscape analysis and model-ing Although the two cities are quite different integra-tion is furthered by a common approach to spatial analy-ses a hierarchical patch dynamics approach (Wu andLoucks 1995 Pickett et al 1999) Hierarchical patchdynamics models start with ecological processes associat-ed with fundamental units of landscapes at some specif-ic scale called patches They then address ecologicalprocesses that are associated with the patches The struc-ture of the patches can be a major determinant of thoseprocesses However patch structure and arrangementalso can change through time Hence models mustaccount for such change that is they must be dynamicFurthermore because patches at a particular scale areoften themselves composed of smaller patches and canbe aggregated into larger patches the models must behierarchical By considering patch dynamics simultane-ously at multiple scales with an accompanying hierarchyof models the complexity of urban systems is rendered

more tractable and translation of information acrossscales is facilitated (Wu 1999)

Special characteristics that dictate a novelframework The modern metropolis presents a striking-ly heterogeneous pattern for study For instance the sharpcontrasts between neighborhoods is a familiar characteris-tic of cities (Clay 1973) Within the span of a city blockwhich is on the order of 200 or fewer meters an observermay cross several obvious boundaries Different kinds ofcommercial use shifts between owner-occupied and rentalproperties and shifts in socioeconomic resources availableto residents are but some of the many contrasts cities pre-sent Such heterogeneity is not unique to dense centralurban districts In fact the contemporary suburb is zonedfor even more discrete transitions than the traditionalmixed use of older cities Residential streets feeder streetscommercial streets strip malls regional malls and indus-trial parks are notable patches in the suburban landscapeOf course the scale of transition in postndashWorld War IIsuburbs tends to be coarser than that of older neighbor-hoods and districts because of the shift to dependence onthe automobile However spatial patchiness in the socialeconomic and infrastructural fabric of metropolitan areasremains their most obvious feature Social scientists havelong been aware of the functional significance of spatial het-erogeneity and mixture of uses within urban areas (Jacobs

578 BioScience July 2000 Vol 50 No 7

Articles

Figure 4 Conceptual scheme adapted to show the interaction among physical ecological engineering social andmanagement variables and drivers in the new Tempe Town Lake Arizona

1962) The ecological significance of such socioeconomicheterogeneity is however an open question Indeed deter-mining to what extent the well-recognized patchiness ofurban areas has ecological dimensions and ecologicalimplications is one of the main motivations of integratedlong-term ecological research in the metropolis

Whatever its ecological significance the conspicuousspatial heterogeneity in urban systems is an entry point forintegration with social science The existence of suchclearly defined patches as neighborhoods and cityscapeswhich combine infrastructural and natural features isapparent to all researchers who must work together togenerate the interdisciplinary synthesis for understandingcities as ecological systems Hydrologists ecologistsdemographers economists engineers and citizens all canand do recognize the spatial heterogeneity of cities Neigh-borhood associations watershed associations censustracts and similar groupings are institutional expressionsof this common recognition Of course each discipline orconstituency may see the boundaries or the most salientfeatures of the patchwork somewhat differently For exam-ple the civil engineer and the urban recreationist will havedifferent views of the boundaries of a watershed The firstmay see a ldquosewershedrdquo the second a visually unified land-scape that is engaging on a morningjog Therefore new multidimensionalclassifications of the heterogeneity ofthe metropolis are required Amongthe principal dimensions of such classi-fications however will be factors thatcontrol the flow of materials energyand information through and withinthe metropolis An emphasis on thesekinds of variables suggests that a spatially explicit ecosys-tem perspective can emerge from the heterogeneity of themetropolis

The interdisciplinary integration required to under-stand the ecological significance of spatial pattern inurban ecosystems must account for several important fea-tures of humans and their institutions Although these fea-tures are implied in the social drivers and phenomena weintroduced earlier (Figure 2 see box page 576) these fea-tures add complexity to ecological studies of the city andtherefore cannot be ignored The spatial heterogeneity ofurban systems is established by formal institutions such aszoning regulations and maintained by other formal insti-tutions such as public works and the courts However lessformal institutions such as families and community asso-ciations also contribute to the spatial structure and itsfunction Humans as individuals and groups are self-aware capable of learning quickly and engaged in exten-sive networks of rapid communication These features ofthe human components of urban systems mean that thefeedback among the biological human infrastructuraland the larger physical contexts can be strong and inmany cases rapid This is one reason that education has

been incorporated into the structure of urban LTER pro-grams We hypothesize that learning about the hetero-geneity and function of an urban area can be a tool thatcitizens and institutions can demonstrably use to improvetheir neighborhoods city and region through manage-ment planning and policy (Grove and Burch 1997)

Patch dynamics and hierarchical approachesSpatial heterogeneity was independently chosen as a start-ing point by both the Central ArizonandashPhoenix and Balti-more Ecosystem Study LTER teams In addition to theadvantages already laid out the patch dynamics approachalso brings the advantage of hierarchical nesting andaggregation Such a flexible hierarchical approach isimportant because the structures and consequently theprocesses that govern the function of the metropolis as anecological system occur at a variety of scales For examplejust because some social processes occur at the scale ofneighborhoods of say 15 square blocks does not meanthat the most important ecological impacts occur at thesame scale Similarly the concentration of resources orwastes in particular patches can be due to decisions madeat great distances from the point of concentration There-fore patches may be scaled up or down for different func-

tional analyses and the configurationof patches that are relevant to specificpaths along which resources and infor-mation flow can be assessed

As an example Baltimore can bedescribed in terms of the five-countymetropolitan area (Figure 5) Within itare three principal watersheds thatextend from the rural hinterlands to the

central city Each of these watersheds can be divided intosubcatchments The 17000-ha Gwynns Falls catchmentcontains 16 smaller watersheds that have been used fordiscussion of management and restoration activities Stillsmaller units can be used for mechanistic studies ofecosystem and socioeconomic processes The fact that dif-ferent nestings are possible within the larger units meansthat the scales at which important interactions occur canbe captured and the promulgation of their effects to dif-ferent scales whether coarser or finer can be determined(see Grove and Burch 1997)

A similar nested set of units is being identified in theCentral ArizonandashPhoenix study area (Figure 5) At thelargest scale three patches exist along an eastndashwest gradi-ent from primarily commercialndashindustrialndashresidential toprimarily agricultural to primarily desert land usendashlandcover The political boundaries of the 24 different munic-ipalities in the Phoenix metropolitan area form anotherset of smaller patches and heterogeneity of land usendashlandcover is evident within these smaller units Interestinglypatch size regularity and connectivity differ within thethree broadest patches (Matthew Luck and Jianguo WuArizona State University personal communication)

July 2000 Vol 50 No 7 BioScience 579

Articles

ldquoThe civil engineer and the

urban recreationist will have

different views of the

boundaries of a watershedrdquo

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

The second class of variables arethose directly associated with humanactivities such as land-use changeintroduction or domestication ofspecies consumption of resourcesand production of wastes

The simplified model shown inFigure 1 defines the intellectual arenawithin which LTER ecologists typi-cally work The inclusion of both nat-ural processes and human activitiesinfluencing long-term ecosystemdynamics is appropriate because evenLTER sites in purportedly pristineareas are subject to human-causeddisturbance and change Patterns andprocesses of ecosystems where thefive core areas (primary productionpopulations organic matter nutri-ents and disturbance) of the LTERprogram are centered have beenmore completely specified than havesocial patterns and processes It is clear however that sev-eral important interactions and feedbacks are missingfrom this approach Because many of these missing fea-tures are the subject matter of the social sciences it isthrough contributions in this area that an understandingof the worldrsquos ecosystems can be most enhanced (Ehrlich1997) At the same time these interactions and feedbacksshould not be pursued in isolation within a self-containedtraditional social-scientific framework because thehuman activities that influence ecosystem dynamics arereciprocally influenced both by biogeophysical drivingforces and by ecosystem dynamics

An example of this reciprocal influence between humanactivity and ecosystem dynamics is provided by work atthe North Temperate Lakes LTER site (Carpenter et al1999) one of two LTER programs that have received addi-tional funding to add social science research and regional-ization to their programs (the other augmented programis the Coweeta LTER) In Wisconsin agricultural land usesare linked to eutrophication of the arearsquos lakes primarilythrough excess phosphorus inputs Farmersrsquo use of fertil-izers can directly affect soil dynamics and soil conditionscan in turn affect a farmerrsquos decision to use fertilizers (seeCarpenter et al 1999) In addition a wide set of socioeco-nomic drivers such as the local or regional economy caninfluence human interaction with the natural landscapeIn the example of fertilizer use the market potential of thecrops government subsidies and even the practices ofneighboring farmers can influence a farmerrsquos decision touse fertilizers (see Carpenter et al 1999) Finally ecosys-tem dynamics themselves as altered by impaired waterquality caused by leaching of excess fertilizer-derivednutrients can influence patterns of socioeconomic activi-ty at a larger scale including real estate values industrial

relocation or recreation patterns Examining these inter-actions both complicates and enhances the long-termstudy of an ecosystem

Without understanding interactions and feedbacksbetween human and ecological systems our view ofecosystem dynamics both at local and global scales will belimitedmdashas will be our ability to apply these insights topublic policy and land management Acknowledging thecentral human component leads to an emphasis on newquantitative methods new approaches to modeling newways to account for risk and value the need to understandenvironmental justice and the importance of workingwithin a globally interacting network (eg Grossmann1993) These added interactions and feedbacks tradition-ally have been studied by social scientists in isolation fromlife and earth scientists Rarely if ever has a focused long-term study incorporated all the interactions implicit inFigure 2 Such integration requires a research team thatbrings together scientists from the natural social andengineering sciences in a unified research endeavor

Ecology in and ecology of urban ecological sys-tems The simplest phrase describing the research beingdone in the Baltimore and Phoenix LTER studies is urbanecology However this term is fraught with misunder-standing because it is often applied to research that doesnot encompass the full suite of concepts we hope to devel-op We prefer to distinguish between two different types ofresearch that more accurately describe the projects weenvision ecology in cities and ecology of cities There areabundant examples in the literature of ecology in cities(eg Sukopp and Werner 1982 Sukopp 1990) The basicquestions addressed by such studies are how do ecologicalpatterns and processes differ in cities as compared with

July 2000 Vol 50 No 7 BioScience 573

Articles

Figure 1 Two classes of variables that affect ecosystems patterns and processes ofecosystems and patterns of human activities Ecologists have mostly studied theformer and have developed theory based on the fundamental biogeophysical driversthat determine ecosystem pattern and process Items listed under ldquopatterns andprocesses of ecosystemsrdquo are the core areas of long-term ecological research

other environments What is the effect of the city (ie aconcentration of human population and activities) on theecology of organisms inside and outside of its boundaryand influence (eg McPherson et al 1997) Research top-ics exemplifying ecology in cities are distribution andabundance of animal and plant populations air pollutionand meteorology patch-specific ecological pattern andprocesses edge effects (because edges are especially pro-nounced in cities) and exoticndashnative species interactionsTools for doing ecology in cities include beforendashafterexperiments which allow the study of effects of rapidchanges occurring in the urban environment and the con-cept of urbanndashrural gradients (eg McDonnell et al1993) which can be a form of space-for-time substitutionto detect impacts of urbanization on ecological processesThese examples help illustrate that ecology has been donein cities for a long time We do not wish to claim that theurban LTERs are the inventors of urban ecological studiesBut perhaps much of the uniqueness of the new urbanLTER programs lies in their attempt to treat cities asecosystems that is to study the ecology of cities

The concept of ecology of cities has to do with how theaggregated parts sum that is how cities process energy ormatter relative to their surroundings Ecologists mighttake many different routes to understanding the ecology ofcities mass balances of nutrients for the entire systempatch dynamics (wherein all patch types and the changesin and among them are considered for the landscape as awhole) ecological effects of land-use change whole-sys-tem metabolism spatial distribution of resources and

populations (eg a metapopulationapproach) and estimation of theecological footprint of a city (Reesand Wackernagel 1994 Folke et al1997) Tools for studying ecology ofcities include the watershedapproach wherein measurement ofinputs and outputs is simplifiedbecause the system is defined as thearea of land drained by a particularstream (eg Bormann and Likens1967 Likens and Bormann 1995)patch dynamics modeling andmonitoring and modeling of land-use change by incorporating remotesensing and GIS (geographic infor-mation systems) methodologies Ofcourse each approach borrows fromthe other and with increasing scalewhat is viewed as the ecology-of-cities approach may become ecologyin an urban patch within the largerregion of which the city is part Ourintent is not to advocate one ap-proach over the other but we doagree with McDonnell and Pickett

(1990 p 1231) who stated nearly a decade ago that ldquostudyof the [city] as an ecosystemwould be a radical expan-sion of ecologyrdquo Thus both the Baltimore and PhoenixLTER programs are doing ecology in cities but are framingtheir work within the context of city as ecosystem

An ecosystem is a piece of earth of any size that containsinteracting biotic and abiotic elements and that interactswith its surroundings By this definition and with Tans-leyrsquos purpose and definition of the term in mindmdashldquothewhole system (in the sense of physics) including not onlythe organism-complex but also the whole complex ofphysical factors forming what we call the environment ofthe biomemdashthe habitat factors in the widest senserdquo (Tans-ley 1935 p 299)mdasha city is most certainly an ecosystemBut few studies have treated cities as such (a notableexception is the study of Hong Kong by Boyden et al 1981see also Boyle et al 1997) Ecosystems have definablestructure and function Structure refers to the componentparts of the system organismal (including human) popu-lations landscape patches soils and geologic parent mate-rial and local atmospheric and hydrologic systemsEcosystem function is a general term referring to the suiteof processes such as primary production ecosystem respi-ration biogeochemical transformations informationtransfer and material transport that occur within ecosys-tems and link the structural components The function ofa whole ecosystem or a part of an ecosystem can bethought of as an integrated measure of what that unit doesin the context of its surroundings For example a compo-nent ecosystem in a region contributes stocks of resources

574 BioScience July 2000 Vol 50 No 7

Articles

Figure 2 A more comprehensive view of the drivers interactions and feedbacksaffecting ecosystem dynamics that recognizes in addition to biogeophysicaldrivers socioeconomic drivers that determine patterns of human activity Itemslisted under ldquopatterns of human activitiesrdquo are suggested core areas of long-termsocial science research

or fluxes of materials to that larger region More specifi-cally a city might contribute airborne particulates ornitrogenous or sulfurous gases to ecosystems locateddownwind of it

In addition to the structure function and processestraditionally studied by ecologists in any ecosystem urbansystems also contain the dominant components of socialinstitutions culture and behavior and the built environ-ment An ecology-in-cities study that incorporates thesecomponents might consider for example how irrigationpractices or creation of hydrologic infrastructure (egcanals pipes or storm drains) influences the distributionof insects on household or neighborhood scales An ecol-ogy-of-cities approach might include models of urbangrowth and spread that reflect economic and social dri-vers for example the tendency of people to want to live onhillslopes (behavior) or the market value of housing neartransportation routes (economics) This type of studymust necessarily involve the reconceptualization ofhuman activities not as disturbances to the ecosystem butas important drivers of and limitations to it (eg Padoch1993) Traditional scholarship of urban systems in thehumanndashsocial ecological geophysical and civil infra-structural domains has been pursued in relative isolationwith each developing its own disciplines methodologiesand evaluation tools (Borden 1993) Urban ecosystemstudies can bring elements of these disparate approachestogether underscoring the interdependence of these phe-nomena

Traditional biological or earth sciencendashbased approach-es to ecosystem studies are insufficient for urban systemsbecause of the interaction of social systems with biogeo-physical systems (Borden 1993) Although some haveargued that humans can be treated as just another animalpopulation albeit an important one (but see Padoch1993) the suite of social drivers of urban ecosystemsmdashinformation flow culture and institutionsmdashare not easilymodeled within a traditional population framework(Padoch 1993 Turner and Meyer 1993) Several modifica-tions of this framework are necessary to successfully inte-grate human activity into an ecological model The first isto acknowledge the primary importance of human deci-sion-making in the dynamics of the urban ecosystem Thisdecision-making operates within a broad context of cul-ture information and institutions This modification putsan appropriate emphasis on the differential creation flowand control of information within the human ecosystemCulture the learned patterns of behavior for each particu-lar society or group and institutions the formal structuresthat codify patterns of behavior also are central compo-nents of decision-making and thus are key to understand-ing environmentally relevant decisions

State variables of urban ecosystems must include morethan measures of population size species diversity andenergy flow They must also include measures of state asperceived by humans often referred to as ldquoquality of liferdquo

Educational opportunities cultural resources recreationwealth aesthetics and community health all are factorsthat may differ among cities yet these variables have fewparallels in traditional ecosystem studies

Special considerations for the human compo-nent of urban ecological systems Ecologists mustask whether theories developed for ecological systems inthe presumed absence of human influence will be appro-priate for systems like cities where human dominance isunquestionable We suspect that simple modification ofecological theory will prove unsatisfactory because themodifications we have just discussed deal with aspects ofhuman social systems that are far from simple Althoughincorporation of existing social science models (Pickett etal 1997) into ecological theory provides a starting pointdevelopment of a new integrative ecology that explicitlyincorporates human decisions culture institutions andeconomic systems will ultimately be needed (JamesCollins Ann Kinzig Nancy Grimm William Fagan DianeHope Jianguo Wu and Elizabeth Borer unpublishedmanuscript) At the same time it is incumbent upon socialscientists who hope for a realistic understanding of theurban system to consider biogeophysical feedbacks andinteractions with the ecological system The suite of socialsystem components that we plan to include in our newmodels is listed in the box on page 576 although not all ofthese components may be relevant to any particular mod-el or process

Within the context of the LTER program the subjectmatter for investigation started with a set of five core areasthat are common to all LTER research primary produc-tion populations representing trophic structure organicmatter storage and dynamics nutrient transport anddynamics and disturbance Given as we have suggested(Figure 1) that three fundamental biogeophysical driversof ecosystems are the flow of energy the flow of informa-tion and the cycling of materials what are the comparabledrivers and core areas of social systems Social scientistsworking at the augmented (Coweeta and North TemperateLakes) and urban LTER sites consider individuals to beguided in their activities by the knowledge beliefs valuesand social resources shared with other members of theirsocial system at different levels of organization (eg atfamily community state and national levels) In this per-spective the three fundamental drivers of human ele-ments in the ecosystem (Figure 2) are

Flow of information and knowledge

Incorporation of culturally based values and perceptions

Creation and maintenance of institutions and organi-zations

These drivers condition human activities and decisions(see also Turner and Meyer 1993) Although understand-ing the nature and interaction of these drivers must be the

July 2000 Vol 50 No 7 BioScience 575

Articles

ultimate goal of most inquiries actual investigations maymore often be oriented toward measuring the patterns ofhuman activity these processes create Defining the fol-lowing core topicsmdasheach characterized by its activitiesstructure and historic trajectorymdashis key to a comprehen-sive approach to human ecosystem analysis (Figure 2)

Demographic pattern

Economic system

Power hierarchy

Land use and management

Designed environment

Collectively these core topics serve as guidelines of inquiryanalogous to the five ecological core areas identified earlyin the LTER programrsquos history On the one hand theyreflect the processes operating within the system on theother they are a practical guide for field investigationsThe entire LTER network (not just the urban and aug-mented sites) thus provides an excellent starting point toincorporate social science research that is relevant to andintegrated with studies of ecosystem change

A conceptual scheme for understanding urbanecological systems We have now identified drivers andpatterns of activity in both ecological and social systemsthat must collectively be considered for a full understand-ing of human-dominated ecosystems A more specificconceptual scheme or model for how a study of urban eco-logical systems can be approached is represented by Figure3 This scheme has been modified through recent discus-sions from schemes presented by the Baltimore and

Phoenix LTER groups in grant proposals and various pub-lications (eg Grimm 1997 Pickett et al 1997) indeed itis still evolving The diagram includes a set of variablesthat are linked by interactions and feedbacks

The two variables in the upper cornersmdashcoarse-scaleenvironmental context and societal patterns and process-esmdashcan be viewed as constraints which are the outcomeof operation of the fundamental biogeophysical and soci-etal drivers The coarser-scale environmental contextincludes climate geology history and biogeographicalsetting and societal patterns and processes encompass thesocioeconomic system culture demography and socialinstitutions

The middle two variables are the focus of the new urbanLTER research land use and ecological patterns andprocesses associated with any given land use Land useincorporates both the intent and the reality of how a giv-en parcel within a metropolitan boundary is altered byhuman decisions whereas ecological patterns and process-esmdashenergy flow nutrient cycling the hydrologic cyclespecies distribution abundance and interactions ecosys-tem and landscape structure and disturbancemdashare thosephenomena studied by all LTER teams Land use and asso-ciated ecological conditions are viewed at a single point intime using a hierarchical patch mosaic perspective Land-use change occurs because of development urban renew-al changes in land management or ownership or infra-structure development among other causes

The last two variables in the lower corners of Figure 3mdashchanges in ecological conditions and changes in humanperceptions and attitudesmdashresult from the interactionbetween land-use change and ecological pattern andprocess Changes in ecological conditions represent thenext time step of ecological patterns and processes andchanges in human perceptions and attitudes represent thehuman reaction to either ecological pattern and process orchanges therein as expressed through the filter of humanexperience

The interactions and feedbacks depicted in Figure 3 areintended to reflect temporal dynamics to some extent Asan example land-use decisions are based on the environ-mental and politicondashsocioeconomic context the land useis then perceived as good or bad and this feedback canlead to different human decisions This sequence of deci-sions and changes in land use does not incorporate newecological information (ie the changes in ecological con-dition that result from the land-use change) such a situa-tion is unstable and seems unlikely to lead to a sustainableurban environment A sequence of interactions and feed-backs carried out when a change occurs or in response toan environmental problem however would incorporateeither short-term solutions to those problems or adjust-ments in management decisions based on a solid ecologi-cal foundation

To illustrate the sequence of changes and feedbacks tofurther change consider an example from the Central

576 BioScience July 2000 Vol 50 No 7

Articles

Social institutions Health Justice Faith Commerce Education Leisure Government

Social dynamics Physiological Individual Organizational Institutional Environmental

Social system componentsThe following are examples of social system componentsto be incorporated into humanndashecological models ofurban ecosystems (from Pickett et al 1997)

Social order Age Gender Class Norms Wealth Power Status Knowledge Territory

Social resources Economic (information

population labor) Cultural (organizations

beliefs myths)

ArizonandashPhoenix LTER sitethe establishment of an artifi-cial lake in the once-dry SaltRiver bed (Figure 4) The ini-tial land-use decision (estab-lishment of the lake) was con-strained on the physicalndashecological side by the existenceof an alluvial channel with nosurface water flow (because ofupstream impoundments) in aregion of North America char-acterized by a high propensityfor flash flooding (Baker1977) Societal constraintsincluded the economic cost ofthe project the perceptions ofpolitical and economic bene-fit available technology (col-lapsible dams recirculatingpumps) and the existence ofhuman-created infrastructure(engineered channel diversion ofsurface flow into canals) Giventhese constraints and based onour best understanding of lakeecology the ecological conditionsassociated with the lake when itwas filled were likely to be highnutrients with concomitant highalgal production high rates ofinfiltration and a high probabilityof floods At the next time step weexpected such changes in ecologi-cal conditions as eutrophicationlosses of water to the groundwatersystem and establishment of arobust mosquito population Pre-liminary monitoring confirms thepredictions of high phytoplank-ton biomass and insect popula-tions and a summer flash floodresulted in a brief episode of high phosphorus loading(Amalfi 1999)

The predicted sequence of ecological conditions mayresult in a full range of reactions from the public Feed-backs to the societal patterns and processesmdashin this casethe design and management decisionsmdashcould lead eitherto chemical additions to control algal blooms or to moreecologically based management practices such as controlof nutrient inputs Indeed complaints from the publicabout insect populations already have resulted in imple-mentation of a chemical control program (Amalfi 1999)and feedbacks of ecological information and knowledge ofthe lakersquos dynamics have helped create and modify a lakemanagement plan Thus societal responses can act directly

on the changed conditions (arrow J in Figure 3 eg addi-tion of chemical control agents the technological solutionin Figure 4) or indirectly via an effect on the underlyingecological patterns and processes producing the problem(arrow K in Figure 3 eg diversion of upstream point-source nutrient inputs to the water supply the arrowlabeled management for sustainability in Figure 4) Webelieve that this general scheme (Figure 3) allows us tothink about how social and ecological systems interact inurban areas In addition to addressing fundamental eco-logical processes this approach allows one to make pre-dictions that are testable and it also provides tools that cit-izens and decision-makers can use to create moreecologically sound policy

July 2000 Vol 50 No 7 BioScience 577

Articles

Figure 3 Conceptual scheme for integrating ecological and social systems in urbanenvironments Variables are in boxes interactions and feedbacks are arrows Aenvironmental context sets the range of possibilities for land usendashland cover B societaldecisions and human behavior (incorporating their suite of determinants) are thedirect drivers of land-use change C the pattern of land use (whatever the driver)determines ecological patterns and processes D humans perceive and react to land-usechange (independent of any ecological effects) E humans also perceive and react toecological patterns and processes F in this interaction ecological processes as affectedby land-use change result in a change in ecological conditions G such changes inecological conditions may result in changes in attitudes (even if human perceptionpreviously ignored ecological pattern and process) and changed ecological conditionsare perceived as good or bad by humans H changes in perception and attitude feedback to the societal system (patterns and processes of society) to influence decision-making and this part of the cycle begins anew I in some cases changed ecologicalconditions can alter the coarse-scale environmental context (example urban heatisland) resulting in a feedback that is relatively independent of human response J KWhen a societal response to changed ecological conditions is deemed necessary thesociety can act directly on the changed conditions (J) or on the underlying ecologicalpatterns and processes producing the problem (K) Finally the environmental contextof course influences ecological patterns independent of land use (L)

A modeling framework for investigating citiesThe conceptual insights of the broad scheme in Figure 3illustrate in general how processes from the social realmbiological environment civil infrastructure and the larg-er climatic and geological context can be integrated withthe specific characteristics of metropolitan Phoenix andmetropolitan Baltimore But the extreme patchiness ofurban environments dictates attention to spatial detailusing novel approaches of landscape analysis and model-ing Although the two cities are quite different integra-tion is furthered by a common approach to spatial analy-ses a hierarchical patch dynamics approach (Wu andLoucks 1995 Pickett et al 1999) Hierarchical patchdynamics models start with ecological processes associat-ed with fundamental units of landscapes at some specif-ic scale called patches They then address ecologicalprocesses that are associated with the patches The struc-ture of the patches can be a major determinant of thoseprocesses However patch structure and arrangementalso can change through time Hence models mustaccount for such change that is they must be dynamicFurthermore because patches at a particular scale areoften themselves composed of smaller patches and canbe aggregated into larger patches the models must behierarchical By considering patch dynamics simultane-ously at multiple scales with an accompanying hierarchyof models the complexity of urban systems is rendered

more tractable and translation of information acrossscales is facilitated (Wu 1999)

Special characteristics that dictate a novelframework The modern metropolis presents a striking-ly heterogeneous pattern for study For instance the sharpcontrasts between neighborhoods is a familiar characteris-tic of cities (Clay 1973) Within the span of a city blockwhich is on the order of 200 or fewer meters an observermay cross several obvious boundaries Different kinds ofcommercial use shifts between owner-occupied and rentalproperties and shifts in socioeconomic resources availableto residents are but some of the many contrasts cities pre-sent Such heterogeneity is not unique to dense centralurban districts In fact the contemporary suburb is zonedfor even more discrete transitions than the traditionalmixed use of older cities Residential streets feeder streetscommercial streets strip malls regional malls and indus-trial parks are notable patches in the suburban landscapeOf course the scale of transition in postndashWorld War IIsuburbs tends to be coarser than that of older neighbor-hoods and districts because of the shift to dependence onthe automobile However spatial patchiness in the socialeconomic and infrastructural fabric of metropolitan areasremains their most obvious feature Social scientists havelong been aware of the functional significance of spatial het-erogeneity and mixture of uses within urban areas (Jacobs

578 BioScience July 2000 Vol 50 No 7

Articles

Figure 4 Conceptual scheme adapted to show the interaction among physical ecological engineering social andmanagement variables and drivers in the new Tempe Town Lake Arizona

1962) The ecological significance of such socioeconomicheterogeneity is however an open question Indeed deter-mining to what extent the well-recognized patchiness ofurban areas has ecological dimensions and ecologicalimplications is one of the main motivations of integratedlong-term ecological research in the metropolis

Whatever its ecological significance the conspicuousspatial heterogeneity in urban systems is an entry point forintegration with social science The existence of suchclearly defined patches as neighborhoods and cityscapeswhich combine infrastructural and natural features isapparent to all researchers who must work together togenerate the interdisciplinary synthesis for understandingcities as ecological systems Hydrologists ecologistsdemographers economists engineers and citizens all canand do recognize the spatial heterogeneity of cities Neigh-borhood associations watershed associations censustracts and similar groupings are institutional expressionsof this common recognition Of course each discipline orconstituency may see the boundaries or the most salientfeatures of the patchwork somewhat differently For exam-ple the civil engineer and the urban recreationist will havedifferent views of the boundaries of a watershed The firstmay see a ldquosewershedrdquo the second a visually unified land-scape that is engaging on a morningjog Therefore new multidimensionalclassifications of the heterogeneity ofthe metropolis are required Amongthe principal dimensions of such classi-fications however will be factors thatcontrol the flow of materials energyand information through and withinthe metropolis An emphasis on thesekinds of variables suggests that a spatially explicit ecosys-tem perspective can emerge from the heterogeneity of themetropolis

The interdisciplinary integration required to under-stand the ecological significance of spatial pattern inurban ecosystems must account for several important fea-tures of humans and their institutions Although these fea-tures are implied in the social drivers and phenomena weintroduced earlier (Figure 2 see box page 576) these fea-tures add complexity to ecological studies of the city andtherefore cannot be ignored The spatial heterogeneity ofurban systems is established by formal institutions such aszoning regulations and maintained by other formal insti-tutions such as public works and the courts However lessformal institutions such as families and community asso-ciations also contribute to the spatial structure and itsfunction Humans as individuals and groups are self-aware capable of learning quickly and engaged in exten-sive networks of rapid communication These features ofthe human components of urban systems mean that thefeedback among the biological human infrastructuraland the larger physical contexts can be strong and inmany cases rapid This is one reason that education has

been incorporated into the structure of urban LTER pro-grams We hypothesize that learning about the hetero-geneity and function of an urban area can be a tool thatcitizens and institutions can demonstrably use to improvetheir neighborhoods city and region through manage-ment planning and policy (Grove and Burch 1997)

Patch dynamics and hierarchical approachesSpatial heterogeneity was independently chosen as a start-ing point by both the Central ArizonandashPhoenix and Balti-more Ecosystem Study LTER teams In addition to theadvantages already laid out the patch dynamics approachalso brings the advantage of hierarchical nesting andaggregation Such a flexible hierarchical approach isimportant because the structures and consequently theprocesses that govern the function of the metropolis as anecological system occur at a variety of scales For examplejust because some social processes occur at the scale ofneighborhoods of say 15 square blocks does not meanthat the most important ecological impacts occur at thesame scale Similarly the concentration of resources orwastes in particular patches can be due to decisions madeat great distances from the point of concentration There-fore patches may be scaled up or down for different func-

tional analyses and the configurationof patches that are relevant to specificpaths along which resources and infor-mation flow can be assessed

As an example Baltimore can bedescribed in terms of the five-countymetropolitan area (Figure 5) Within itare three principal watersheds thatextend from the rural hinterlands to the

central city Each of these watersheds can be divided intosubcatchments The 17000-ha Gwynns Falls catchmentcontains 16 smaller watersheds that have been used fordiscussion of management and restoration activities Stillsmaller units can be used for mechanistic studies ofecosystem and socioeconomic processes The fact that dif-ferent nestings are possible within the larger units meansthat the scales at which important interactions occur canbe captured and the promulgation of their effects to dif-ferent scales whether coarser or finer can be determined(see Grove and Burch 1997)

A similar nested set of units is being identified in theCentral ArizonandashPhoenix study area (Figure 5) At thelargest scale three patches exist along an eastndashwest gradi-ent from primarily commercialndashindustrialndashresidential toprimarily agricultural to primarily desert land usendashlandcover The political boundaries of the 24 different munic-ipalities in the Phoenix metropolitan area form anotherset of smaller patches and heterogeneity of land usendashlandcover is evident within these smaller units Interestinglypatch size regularity and connectivity differ within thethree broadest patches (Matthew Luck and Jianguo WuArizona State University personal communication)

July 2000 Vol 50 No 7 BioScience 579

Articles

ldquoThe civil engineer and the

urban recreationist will have

different views of the

boundaries of a watershedrdquo

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

other environments What is the effect of the city (ie aconcentration of human population and activities) on theecology of organisms inside and outside of its boundaryand influence (eg McPherson et al 1997) Research top-ics exemplifying ecology in cities are distribution andabundance of animal and plant populations air pollutionand meteorology patch-specific ecological pattern andprocesses edge effects (because edges are especially pro-nounced in cities) and exoticndashnative species interactionsTools for doing ecology in cities include beforendashafterexperiments which allow the study of effects of rapidchanges occurring in the urban environment and the con-cept of urbanndashrural gradients (eg McDonnell et al1993) which can be a form of space-for-time substitutionto detect impacts of urbanization on ecological processesThese examples help illustrate that ecology has been donein cities for a long time We do not wish to claim that theurban LTERs are the inventors of urban ecological studiesBut perhaps much of the uniqueness of the new urbanLTER programs lies in their attempt to treat cities asecosystems that is to study the ecology of cities

The concept of ecology of cities has to do with how theaggregated parts sum that is how cities process energy ormatter relative to their surroundings Ecologists mighttake many different routes to understanding the ecology ofcities mass balances of nutrients for the entire systempatch dynamics (wherein all patch types and the changesin and among them are considered for the landscape as awhole) ecological effects of land-use change whole-sys-tem metabolism spatial distribution of resources and

populations (eg a metapopulationapproach) and estimation of theecological footprint of a city (Reesand Wackernagel 1994 Folke et al1997) Tools for studying ecology ofcities include the watershedapproach wherein measurement ofinputs and outputs is simplifiedbecause the system is defined as thearea of land drained by a particularstream (eg Bormann and Likens1967 Likens and Bormann 1995)patch dynamics modeling andmonitoring and modeling of land-use change by incorporating remotesensing and GIS (geographic infor-mation systems) methodologies Ofcourse each approach borrows fromthe other and with increasing scalewhat is viewed as the ecology-of-cities approach may become ecologyin an urban patch within the largerregion of which the city is part Ourintent is not to advocate one ap-proach over the other but we doagree with McDonnell and Pickett

(1990 p 1231) who stated nearly a decade ago that ldquostudyof the [city] as an ecosystemwould be a radical expan-sion of ecologyrdquo Thus both the Baltimore and PhoenixLTER programs are doing ecology in cities but are framingtheir work within the context of city as ecosystem

An ecosystem is a piece of earth of any size that containsinteracting biotic and abiotic elements and that interactswith its surroundings By this definition and with Tans-leyrsquos purpose and definition of the term in mindmdashldquothewhole system (in the sense of physics) including not onlythe organism-complex but also the whole complex ofphysical factors forming what we call the environment ofthe biomemdashthe habitat factors in the widest senserdquo (Tans-ley 1935 p 299)mdasha city is most certainly an ecosystemBut few studies have treated cities as such (a notableexception is the study of Hong Kong by Boyden et al 1981see also Boyle et al 1997) Ecosystems have definablestructure and function Structure refers to the componentparts of the system organismal (including human) popu-lations landscape patches soils and geologic parent mate-rial and local atmospheric and hydrologic systemsEcosystem function is a general term referring to the suiteof processes such as primary production ecosystem respi-ration biogeochemical transformations informationtransfer and material transport that occur within ecosys-tems and link the structural components The function ofa whole ecosystem or a part of an ecosystem can bethought of as an integrated measure of what that unit doesin the context of its surroundings For example a compo-nent ecosystem in a region contributes stocks of resources

574 BioScience July 2000 Vol 50 No 7

Articles

Figure 2 A more comprehensive view of the drivers interactions and feedbacksaffecting ecosystem dynamics that recognizes in addition to biogeophysicaldrivers socioeconomic drivers that determine patterns of human activity Itemslisted under ldquopatterns of human activitiesrdquo are suggested core areas of long-termsocial science research

or fluxes of materials to that larger region More specifi-cally a city might contribute airborne particulates ornitrogenous or sulfurous gases to ecosystems locateddownwind of it

In addition to the structure function and processestraditionally studied by ecologists in any ecosystem urbansystems also contain the dominant components of socialinstitutions culture and behavior and the built environ-ment An ecology-in-cities study that incorporates thesecomponents might consider for example how irrigationpractices or creation of hydrologic infrastructure (egcanals pipes or storm drains) influences the distributionof insects on household or neighborhood scales An ecol-ogy-of-cities approach might include models of urbangrowth and spread that reflect economic and social dri-vers for example the tendency of people to want to live onhillslopes (behavior) or the market value of housing neartransportation routes (economics) This type of studymust necessarily involve the reconceptualization ofhuman activities not as disturbances to the ecosystem butas important drivers of and limitations to it (eg Padoch1993) Traditional scholarship of urban systems in thehumanndashsocial ecological geophysical and civil infra-structural domains has been pursued in relative isolationwith each developing its own disciplines methodologiesand evaluation tools (Borden 1993) Urban ecosystemstudies can bring elements of these disparate approachestogether underscoring the interdependence of these phe-nomena

Traditional biological or earth sciencendashbased approach-es to ecosystem studies are insufficient for urban systemsbecause of the interaction of social systems with biogeo-physical systems (Borden 1993) Although some haveargued that humans can be treated as just another animalpopulation albeit an important one (but see Padoch1993) the suite of social drivers of urban ecosystemsmdashinformation flow culture and institutionsmdashare not easilymodeled within a traditional population framework(Padoch 1993 Turner and Meyer 1993) Several modifica-tions of this framework are necessary to successfully inte-grate human activity into an ecological model The first isto acknowledge the primary importance of human deci-sion-making in the dynamics of the urban ecosystem Thisdecision-making operates within a broad context of cul-ture information and institutions This modification putsan appropriate emphasis on the differential creation flowand control of information within the human ecosystemCulture the learned patterns of behavior for each particu-lar society or group and institutions the formal structuresthat codify patterns of behavior also are central compo-nents of decision-making and thus are key to understand-ing environmentally relevant decisions

State variables of urban ecosystems must include morethan measures of population size species diversity andenergy flow They must also include measures of state asperceived by humans often referred to as ldquoquality of liferdquo

Educational opportunities cultural resources recreationwealth aesthetics and community health all are factorsthat may differ among cities yet these variables have fewparallels in traditional ecosystem studies

Special considerations for the human compo-nent of urban ecological systems Ecologists mustask whether theories developed for ecological systems inthe presumed absence of human influence will be appro-priate for systems like cities where human dominance isunquestionable We suspect that simple modification ofecological theory will prove unsatisfactory because themodifications we have just discussed deal with aspects ofhuman social systems that are far from simple Althoughincorporation of existing social science models (Pickett etal 1997) into ecological theory provides a starting pointdevelopment of a new integrative ecology that explicitlyincorporates human decisions culture institutions andeconomic systems will ultimately be needed (JamesCollins Ann Kinzig Nancy Grimm William Fagan DianeHope Jianguo Wu and Elizabeth Borer unpublishedmanuscript) At the same time it is incumbent upon socialscientists who hope for a realistic understanding of theurban system to consider biogeophysical feedbacks andinteractions with the ecological system The suite of socialsystem components that we plan to include in our newmodels is listed in the box on page 576 although not all ofthese components may be relevant to any particular mod-el or process

Within the context of the LTER program the subjectmatter for investigation started with a set of five core areasthat are common to all LTER research primary produc-tion populations representing trophic structure organicmatter storage and dynamics nutrient transport anddynamics and disturbance Given as we have suggested(Figure 1) that three fundamental biogeophysical driversof ecosystems are the flow of energy the flow of informa-tion and the cycling of materials what are the comparabledrivers and core areas of social systems Social scientistsworking at the augmented (Coweeta and North TemperateLakes) and urban LTER sites consider individuals to beguided in their activities by the knowledge beliefs valuesand social resources shared with other members of theirsocial system at different levels of organization (eg atfamily community state and national levels) In this per-spective the three fundamental drivers of human ele-ments in the ecosystem (Figure 2) are

Flow of information and knowledge

Incorporation of culturally based values and perceptions

Creation and maintenance of institutions and organi-zations

These drivers condition human activities and decisions(see also Turner and Meyer 1993) Although understand-ing the nature and interaction of these drivers must be the

July 2000 Vol 50 No 7 BioScience 575

Articles

ultimate goal of most inquiries actual investigations maymore often be oriented toward measuring the patterns ofhuman activity these processes create Defining the fol-lowing core topicsmdasheach characterized by its activitiesstructure and historic trajectorymdashis key to a comprehen-sive approach to human ecosystem analysis (Figure 2)

Demographic pattern

Economic system

Power hierarchy

Land use and management

Designed environment

Collectively these core topics serve as guidelines of inquiryanalogous to the five ecological core areas identified earlyin the LTER programrsquos history On the one hand theyreflect the processes operating within the system on theother they are a practical guide for field investigationsThe entire LTER network (not just the urban and aug-mented sites) thus provides an excellent starting point toincorporate social science research that is relevant to andintegrated with studies of ecosystem change

A conceptual scheme for understanding urbanecological systems We have now identified drivers andpatterns of activity in both ecological and social systemsthat must collectively be considered for a full understand-ing of human-dominated ecosystems A more specificconceptual scheme or model for how a study of urban eco-logical systems can be approached is represented by Figure3 This scheme has been modified through recent discus-sions from schemes presented by the Baltimore and

Phoenix LTER groups in grant proposals and various pub-lications (eg Grimm 1997 Pickett et al 1997) indeed itis still evolving The diagram includes a set of variablesthat are linked by interactions and feedbacks

The two variables in the upper cornersmdashcoarse-scaleenvironmental context and societal patterns and process-esmdashcan be viewed as constraints which are the outcomeof operation of the fundamental biogeophysical and soci-etal drivers The coarser-scale environmental contextincludes climate geology history and biogeographicalsetting and societal patterns and processes encompass thesocioeconomic system culture demography and socialinstitutions

The middle two variables are the focus of the new urbanLTER research land use and ecological patterns andprocesses associated with any given land use Land useincorporates both the intent and the reality of how a giv-en parcel within a metropolitan boundary is altered byhuman decisions whereas ecological patterns and process-esmdashenergy flow nutrient cycling the hydrologic cyclespecies distribution abundance and interactions ecosys-tem and landscape structure and disturbancemdashare thosephenomena studied by all LTER teams Land use and asso-ciated ecological conditions are viewed at a single point intime using a hierarchical patch mosaic perspective Land-use change occurs because of development urban renew-al changes in land management or ownership or infra-structure development among other causes

The last two variables in the lower corners of Figure 3mdashchanges in ecological conditions and changes in humanperceptions and attitudesmdashresult from the interactionbetween land-use change and ecological pattern andprocess Changes in ecological conditions represent thenext time step of ecological patterns and processes andchanges in human perceptions and attitudes represent thehuman reaction to either ecological pattern and process orchanges therein as expressed through the filter of humanexperience

The interactions and feedbacks depicted in Figure 3 areintended to reflect temporal dynamics to some extent Asan example land-use decisions are based on the environ-mental and politicondashsocioeconomic context the land useis then perceived as good or bad and this feedback canlead to different human decisions This sequence of deci-sions and changes in land use does not incorporate newecological information (ie the changes in ecological con-dition that result from the land-use change) such a situa-tion is unstable and seems unlikely to lead to a sustainableurban environment A sequence of interactions and feed-backs carried out when a change occurs or in response toan environmental problem however would incorporateeither short-term solutions to those problems or adjust-ments in management decisions based on a solid ecologi-cal foundation

To illustrate the sequence of changes and feedbacks tofurther change consider an example from the Central

576 BioScience July 2000 Vol 50 No 7

Articles

Social institutions Health Justice Faith Commerce Education Leisure Government

Social dynamics Physiological Individual Organizational Institutional Environmental

Social system componentsThe following are examples of social system componentsto be incorporated into humanndashecological models ofurban ecosystems (from Pickett et al 1997)

Social order Age Gender Class Norms Wealth Power Status Knowledge Territory

Social resources Economic (information

population labor) Cultural (organizations

beliefs myths)

ArizonandashPhoenix LTER sitethe establishment of an artifi-cial lake in the once-dry SaltRiver bed (Figure 4) The ini-tial land-use decision (estab-lishment of the lake) was con-strained on the physicalndashecological side by the existenceof an alluvial channel with nosurface water flow (because ofupstream impoundments) in aregion of North America char-acterized by a high propensityfor flash flooding (Baker1977) Societal constraintsincluded the economic cost ofthe project the perceptions ofpolitical and economic bene-fit available technology (col-lapsible dams recirculatingpumps) and the existence ofhuman-created infrastructure(engineered channel diversion ofsurface flow into canals) Giventhese constraints and based onour best understanding of lakeecology the ecological conditionsassociated with the lake when itwas filled were likely to be highnutrients with concomitant highalgal production high rates ofinfiltration and a high probabilityof floods At the next time step weexpected such changes in ecologi-cal conditions as eutrophicationlosses of water to the groundwatersystem and establishment of arobust mosquito population Pre-liminary monitoring confirms thepredictions of high phytoplank-ton biomass and insect popula-tions and a summer flash floodresulted in a brief episode of high phosphorus loading(Amalfi 1999)

The predicted sequence of ecological conditions mayresult in a full range of reactions from the public Feed-backs to the societal patterns and processesmdashin this casethe design and management decisionsmdashcould lead eitherto chemical additions to control algal blooms or to moreecologically based management practices such as controlof nutrient inputs Indeed complaints from the publicabout insect populations already have resulted in imple-mentation of a chemical control program (Amalfi 1999)and feedbacks of ecological information and knowledge ofthe lakersquos dynamics have helped create and modify a lakemanagement plan Thus societal responses can act directly

on the changed conditions (arrow J in Figure 3 eg addi-tion of chemical control agents the technological solutionin Figure 4) or indirectly via an effect on the underlyingecological patterns and processes producing the problem(arrow K in Figure 3 eg diversion of upstream point-source nutrient inputs to the water supply the arrowlabeled management for sustainability in Figure 4) Webelieve that this general scheme (Figure 3) allows us tothink about how social and ecological systems interact inurban areas In addition to addressing fundamental eco-logical processes this approach allows one to make pre-dictions that are testable and it also provides tools that cit-izens and decision-makers can use to create moreecologically sound policy

July 2000 Vol 50 No 7 BioScience 577

Articles

Figure 3 Conceptual scheme for integrating ecological and social systems in urbanenvironments Variables are in boxes interactions and feedbacks are arrows Aenvironmental context sets the range of possibilities for land usendashland cover B societaldecisions and human behavior (incorporating their suite of determinants) are thedirect drivers of land-use change C the pattern of land use (whatever the driver)determines ecological patterns and processes D humans perceive and react to land-usechange (independent of any ecological effects) E humans also perceive and react toecological patterns and processes F in this interaction ecological processes as affectedby land-use change result in a change in ecological conditions G such changes inecological conditions may result in changes in attitudes (even if human perceptionpreviously ignored ecological pattern and process) and changed ecological conditionsare perceived as good or bad by humans H changes in perception and attitude feedback to the societal system (patterns and processes of society) to influence decision-making and this part of the cycle begins anew I in some cases changed ecologicalconditions can alter the coarse-scale environmental context (example urban heatisland) resulting in a feedback that is relatively independent of human response J KWhen a societal response to changed ecological conditions is deemed necessary thesociety can act directly on the changed conditions (J) or on the underlying ecologicalpatterns and processes producing the problem (K) Finally the environmental contextof course influences ecological patterns independent of land use (L)

A modeling framework for investigating citiesThe conceptual insights of the broad scheme in Figure 3illustrate in general how processes from the social realmbiological environment civil infrastructure and the larg-er climatic and geological context can be integrated withthe specific characteristics of metropolitan Phoenix andmetropolitan Baltimore But the extreme patchiness ofurban environments dictates attention to spatial detailusing novel approaches of landscape analysis and model-ing Although the two cities are quite different integra-tion is furthered by a common approach to spatial analy-ses a hierarchical patch dynamics approach (Wu andLoucks 1995 Pickett et al 1999) Hierarchical patchdynamics models start with ecological processes associat-ed with fundamental units of landscapes at some specif-ic scale called patches They then address ecologicalprocesses that are associated with the patches The struc-ture of the patches can be a major determinant of thoseprocesses However patch structure and arrangementalso can change through time Hence models mustaccount for such change that is they must be dynamicFurthermore because patches at a particular scale areoften themselves composed of smaller patches and canbe aggregated into larger patches the models must behierarchical By considering patch dynamics simultane-ously at multiple scales with an accompanying hierarchyof models the complexity of urban systems is rendered

more tractable and translation of information acrossscales is facilitated (Wu 1999)

Special characteristics that dictate a novelframework The modern metropolis presents a striking-ly heterogeneous pattern for study For instance the sharpcontrasts between neighborhoods is a familiar characteris-tic of cities (Clay 1973) Within the span of a city blockwhich is on the order of 200 or fewer meters an observermay cross several obvious boundaries Different kinds ofcommercial use shifts between owner-occupied and rentalproperties and shifts in socioeconomic resources availableto residents are but some of the many contrasts cities pre-sent Such heterogeneity is not unique to dense centralurban districts In fact the contemporary suburb is zonedfor even more discrete transitions than the traditionalmixed use of older cities Residential streets feeder streetscommercial streets strip malls regional malls and indus-trial parks are notable patches in the suburban landscapeOf course the scale of transition in postndashWorld War IIsuburbs tends to be coarser than that of older neighbor-hoods and districts because of the shift to dependence onthe automobile However spatial patchiness in the socialeconomic and infrastructural fabric of metropolitan areasremains their most obvious feature Social scientists havelong been aware of the functional significance of spatial het-erogeneity and mixture of uses within urban areas (Jacobs

578 BioScience July 2000 Vol 50 No 7

Articles

Figure 4 Conceptual scheme adapted to show the interaction among physical ecological engineering social andmanagement variables and drivers in the new Tempe Town Lake Arizona

1962) The ecological significance of such socioeconomicheterogeneity is however an open question Indeed deter-mining to what extent the well-recognized patchiness ofurban areas has ecological dimensions and ecologicalimplications is one of the main motivations of integratedlong-term ecological research in the metropolis

Whatever its ecological significance the conspicuousspatial heterogeneity in urban systems is an entry point forintegration with social science The existence of suchclearly defined patches as neighborhoods and cityscapeswhich combine infrastructural and natural features isapparent to all researchers who must work together togenerate the interdisciplinary synthesis for understandingcities as ecological systems Hydrologists ecologistsdemographers economists engineers and citizens all canand do recognize the spatial heterogeneity of cities Neigh-borhood associations watershed associations censustracts and similar groupings are institutional expressionsof this common recognition Of course each discipline orconstituency may see the boundaries or the most salientfeatures of the patchwork somewhat differently For exam-ple the civil engineer and the urban recreationist will havedifferent views of the boundaries of a watershed The firstmay see a ldquosewershedrdquo the second a visually unified land-scape that is engaging on a morningjog Therefore new multidimensionalclassifications of the heterogeneity ofthe metropolis are required Amongthe principal dimensions of such classi-fications however will be factors thatcontrol the flow of materials energyand information through and withinthe metropolis An emphasis on thesekinds of variables suggests that a spatially explicit ecosys-tem perspective can emerge from the heterogeneity of themetropolis

The interdisciplinary integration required to under-stand the ecological significance of spatial pattern inurban ecosystems must account for several important fea-tures of humans and their institutions Although these fea-tures are implied in the social drivers and phenomena weintroduced earlier (Figure 2 see box page 576) these fea-tures add complexity to ecological studies of the city andtherefore cannot be ignored The spatial heterogeneity ofurban systems is established by formal institutions such aszoning regulations and maintained by other formal insti-tutions such as public works and the courts However lessformal institutions such as families and community asso-ciations also contribute to the spatial structure and itsfunction Humans as individuals and groups are self-aware capable of learning quickly and engaged in exten-sive networks of rapid communication These features ofthe human components of urban systems mean that thefeedback among the biological human infrastructuraland the larger physical contexts can be strong and inmany cases rapid This is one reason that education has

been incorporated into the structure of urban LTER pro-grams We hypothesize that learning about the hetero-geneity and function of an urban area can be a tool thatcitizens and institutions can demonstrably use to improvetheir neighborhoods city and region through manage-ment planning and policy (Grove and Burch 1997)

Patch dynamics and hierarchical approachesSpatial heterogeneity was independently chosen as a start-ing point by both the Central ArizonandashPhoenix and Balti-more Ecosystem Study LTER teams In addition to theadvantages already laid out the patch dynamics approachalso brings the advantage of hierarchical nesting andaggregation Such a flexible hierarchical approach isimportant because the structures and consequently theprocesses that govern the function of the metropolis as anecological system occur at a variety of scales For examplejust because some social processes occur at the scale ofneighborhoods of say 15 square blocks does not meanthat the most important ecological impacts occur at thesame scale Similarly the concentration of resources orwastes in particular patches can be due to decisions madeat great distances from the point of concentration There-fore patches may be scaled up or down for different func-

tional analyses and the configurationof patches that are relevant to specificpaths along which resources and infor-mation flow can be assessed

As an example Baltimore can bedescribed in terms of the five-countymetropolitan area (Figure 5) Within itare three principal watersheds thatextend from the rural hinterlands to the

central city Each of these watersheds can be divided intosubcatchments The 17000-ha Gwynns Falls catchmentcontains 16 smaller watersheds that have been used fordiscussion of management and restoration activities Stillsmaller units can be used for mechanistic studies ofecosystem and socioeconomic processes The fact that dif-ferent nestings are possible within the larger units meansthat the scales at which important interactions occur canbe captured and the promulgation of their effects to dif-ferent scales whether coarser or finer can be determined(see Grove and Burch 1997)

A similar nested set of units is being identified in theCentral ArizonandashPhoenix study area (Figure 5) At thelargest scale three patches exist along an eastndashwest gradi-ent from primarily commercialndashindustrialndashresidential toprimarily agricultural to primarily desert land usendashlandcover The political boundaries of the 24 different munic-ipalities in the Phoenix metropolitan area form anotherset of smaller patches and heterogeneity of land usendashlandcover is evident within these smaller units Interestinglypatch size regularity and connectivity differ within thethree broadest patches (Matthew Luck and Jianguo WuArizona State University personal communication)

July 2000 Vol 50 No 7 BioScience 579

Articles

ldquoThe civil engineer and the

urban recreationist will have

different views of the

boundaries of a watershedrdquo

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

or fluxes of materials to that larger region More specifi-cally a city might contribute airborne particulates ornitrogenous or sulfurous gases to ecosystems locateddownwind of it

In addition to the structure function and processestraditionally studied by ecologists in any ecosystem urbansystems also contain the dominant components of socialinstitutions culture and behavior and the built environ-ment An ecology-in-cities study that incorporates thesecomponents might consider for example how irrigationpractices or creation of hydrologic infrastructure (egcanals pipes or storm drains) influences the distributionof insects on household or neighborhood scales An ecol-ogy-of-cities approach might include models of urbangrowth and spread that reflect economic and social dri-vers for example the tendency of people to want to live onhillslopes (behavior) or the market value of housing neartransportation routes (economics) This type of studymust necessarily involve the reconceptualization ofhuman activities not as disturbances to the ecosystem butas important drivers of and limitations to it (eg Padoch1993) Traditional scholarship of urban systems in thehumanndashsocial ecological geophysical and civil infra-structural domains has been pursued in relative isolationwith each developing its own disciplines methodologiesand evaluation tools (Borden 1993) Urban ecosystemstudies can bring elements of these disparate approachestogether underscoring the interdependence of these phe-nomena

Traditional biological or earth sciencendashbased approach-es to ecosystem studies are insufficient for urban systemsbecause of the interaction of social systems with biogeo-physical systems (Borden 1993) Although some haveargued that humans can be treated as just another animalpopulation albeit an important one (but see Padoch1993) the suite of social drivers of urban ecosystemsmdashinformation flow culture and institutionsmdashare not easilymodeled within a traditional population framework(Padoch 1993 Turner and Meyer 1993) Several modifica-tions of this framework are necessary to successfully inte-grate human activity into an ecological model The first isto acknowledge the primary importance of human deci-sion-making in the dynamics of the urban ecosystem Thisdecision-making operates within a broad context of cul-ture information and institutions This modification putsan appropriate emphasis on the differential creation flowand control of information within the human ecosystemCulture the learned patterns of behavior for each particu-lar society or group and institutions the formal structuresthat codify patterns of behavior also are central compo-nents of decision-making and thus are key to understand-ing environmentally relevant decisions

State variables of urban ecosystems must include morethan measures of population size species diversity andenergy flow They must also include measures of state asperceived by humans often referred to as ldquoquality of liferdquo

Educational opportunities cultural resources recreationwealth aesthetics and community health all are factorsthat may differ among cities yet these variables have fewparallels in traditional ecosystem studies

Special considerations for the human compo-nent of urban ecological systems Ecologists mustask whether theories developed for ecological systems inthe presumed absence of human influence will be appro-priate for systems like cities where human dominance isunquestionable We suspect that simple modification ofecological theory will prove unsatisfactory because themodifications we have just discussed deal with aspects ofhuman social systems that are far from simple Althoughincorporation of existing social science models (Pickett etal 1997) into ecological theory provides a starting pointdevelopment of a new integrative ecology that explicitlyincorporates human decisions culture institutions andeconomic systems will ultimately be needed (JamesCollins Ann Kinzig Nancy Grimm William Fagan DianeHope Jianguo Wu and Elizabeth Borer unpublishedmanuscript) At the same time it is incumbent upon socialscientists who hope for a realistic understanding of theurban system to consider biogeophysical feedbacks andinteractions with the ecological system The suite of socialsystem components that we plan to include in our newmodels is listed in the box on page 576 although not all ofthese components may be relevant to any particular mod-el or process

Within the context of the LTER program the subjectmatter for investigation started with a set of five core areasthat are common to all LTER research primary produc-tion populations representing trophic structure organicmatter storage and dynamics nutrient transport anddynamics and disturbance Given as we have suggested(Figure 1) that three fundamental biogeophysical driversof ecosystems are the flow of energy the flow of informa-tion and the cycling of materials what are the comparabledrivers and core areas of social systems Social scientistsworking at the augmented (Coweeta and North TemperateLakes) and urban LTER sites consider individuals to beguided in their activities by the knowledge beliefs valuesand social resources shared with other members of theirsocial system at different levels of organization (eg atfamily community state and national levels) In this per-spective the three fundamental drivers of human ele-ments in the ecosystem (Figure 2) are

Flow of information and knowledge

Incorporation of culturally based values and perceptions

Creation and maintenance of institutions and organi-zations

These drivers condition human activities and decisions(see also Turner and Meyer 1993) Although understand-ing the nature and interaction of these drivers must be the

July 2000 Vol 50 No 7 BioScience 575

Articles

ultimate goal of most inquiries actual investigations maymore often be oriented toward measuring the patterns ofhuman activity these processes create Defining the fol-lowing core topicsmdasheach characterized by its activitiesstructure and historic trajectorymdashis key to a comprehen-sive approach to human ecosystem analysis (Figure 2)

Demographic pattern

Economic system

Power hierarchy

Land use and management

Designed environment

Collectively these core topics serve as guidelines of inquiryanalogous to the five ecological core areas identified earlyin the LTER programrsquos history On the one hand theyreflect the processes operating within the system on theother they are a practical guide for field investigationsThe entire LTER network (not just the urban and aug-mented sites) thus provides an excellent starting point toincorporate social science research that is relevant to andintegrated with studies of ecosystem change

A conceptual scheme for understanding urbanecological systems We have now identified drivers andpatterns of activity in both ecological and social systemsthat must collectively be considered for a full understand-ing of human-dominated ecosystems A more specificconceptual scheme or model for how a study of urban eco-logical systems can be approached is represented by Figure3 This scheme has been modified through recent discus-sions from schemes presented by the Baltimore and

Phoenix LTER groups in grant proposals and various pub-lications (eg Grimm 1997 Pickett et al 1997) indeed itis still evolving The diagram includes a set of variablesthat are linked by interactions and feedbacks

The two variables in the upper cornersmdashcoarse-scaleenvironmental context and societal patterns and process-esmdashcan be viewed as constraints which are the outcomeof operation of the fundamental biogeophysical and soci-etal drivers The coarser-scale environmental contextincludes climate geology history and biogeographicalsetting and societal patterns and processes encompass thesocioeconomic system culture demography and socialinstitutions

The middle two variables are the focus of the new urbanLTER research land use and ecological patterns andprocesses associated with any given land use Land useincorporates both the intent and the reality of how a giv-en parcel within a metropolitan boundary is altered byhuman decisions whereas ecological patterns and process-esmdashenergy flow nutrient cycling the hydrologic cyclespecies distribution abundance and interactions ecosys-tem and landscape structure and disturbancemdashare thosephenomena studied by all LTER teams Land use and asso-ciated ecological conditions are viewed at a single point intime using a hierarchical patch mosaic perspective Land-use change occurs because of development urban renew-al changes in land management or ownership or infra-structure development among other causes

The last two variables in the lower corners of Figure 3mdashchanges in ecological conditions and changes in humanperceptions and attitudesmdashresult from the interactionbetween land-use change and ecological pattern andprocess Changes in ecological conditions represent thenext time step of ecological patterns and processes andchanges in human perceptions and attitudes represent thehuman reaction to either ecological pattern and process orchanges therein as expressed through the filter of humanexperience

The interactions and feedbacks depicted in Figure 3 areintended to reflect temporal dynamics to some extent Asan example land-use decisions are based on the environ-mental and politicondashsocioeconomic context the land useis then perceived as good or bad and this feedback canlead to different human decisions This sequence of deci-sions and changes in land use does not incorporate newecological information (ie the changes in ecological con-dition that result from the land-use change) such a situa-tion is unstable and seems unlikely to lead to a sustainableurban environment A sequence of interactions and feed-backs carried out when a change occurs or in response toan environmental problem however would incorporateeither short-term solutions to those problems or adjust-ments in management decisions based on a solid ecologi-cal foundation

To illustrate the sequence of changes and feedbacks tofurther change consider an example from the Central

576 BioScience July 2000 Vol 50 No 7

Articles

Social institutions Health Justice Faith Commerce Education Leisure Government

Social dynamics Physiological Individual Organizational Institutional Environmental

Social system componentsThe following are examples of social system componentsto be incorporated into humanndashecological models ofurban ecosystems (from Pickett et al 1997)

Social order Age Gender Class Norms Wealth Power Status Knowledge Territory

Social resources Economic (information

population labor) Cultural (organizations

beliefs myths)

ArizonandashPhoenix LTER sitethe establishment of an artifi-cial lake in the once-dry SaltRiver bed (Figure 4) The ini-tial land-use decision (estab-lishment of the lake) was con-strained on the physicalndashecological side by the existenceof an alluvial channel with nosurface water flow (because ofupstream impoundments) in aregion of North America char-acterized by a high propensityfor flash flooding (Baker1977) Societal constraintsincluded the economic cost ofthe project the perceptions ofpolitical and economic bene-fit available technology (col-lapsible dams recirculatingpumps) and the existence ofhuman-created infrastructure(engineered channel diversion ofsurface flow into canals) Giventhese constraints and based onour best understanding of lakeecology the ecological conditionsassociated with the lake when itwas filled were likely to be highnutrients with concomitant highalgal production high rates ofinfiltration and a high probabilityof floods At the next time step weexpected such changes in ecologi-cal conditions as eutrophicationlosses of water to the groundwatersystem and establishment of arobust mosquito population Pre-liminary monitoring confirms thepredictions of high phytoplank-ton biomass and insect popula-tions and a summer flash floodresulted in a brief episode of high phosphorus loading(Amalfi 1999)

The predicted sequence of ecological conditions mayresult in a full range of reactions from the public Feed-backs to the societal patterns and processesmdashin this casethe design and management decisionsmdashcould lead eitherto chemical additions to control algal blooms or to moreecologically based management practices such as controlof nutrient inputs Indeed complaints from the publicabout insect populations already have resulted in imple-mentation of a chemical control program (Amalfi 1999)and feedbacks of ecological information and knowledge ofthe lakersquos dynamics have helped create and modify a lakemanagement plan Thus societal responses can act directly

on the changed conditions (arrow J in Figure 3 eg addi-tion of chemical control agents the technological solutionin Figure 4) or indirectly via an effect on the underlyingecological patterns and processes producing the problem(arrow K in Figure 3 eg diversion of upstream point-source nutrient inputs to the water supply the arrowlabeled management for sustainability in Figure 4) Webelieve that this general scheme (Figure 3) allows us tothink about how social and ecological systems interact inurban areas In addition to addressing fundamental eco-logical processes this approach allows one to make pre-dictions that are testable and it also provides tools that cit-izens and decision-makers can use to create moreecologically sound policy

July 2000 Vol 50 No 7 BioScience 577

Articles

Figure 3 Conceptual scheme for integrating ecological and social systems in urbanenvironments Variables are in boxes interactions and feedbacks are arrows Aenvironmental context sets the range of possibilities for land usendashland cover B societaldecisions and human behavior (incorporating their suite of determinants) are thedirect drivers of land-use change C the pattern of land use (whatever the driver)determines ecological patterns and processes D humans perceive and react to land-usechange (independent of any ecological effects) E humans also perceive and react toecological patterns and processes F in this interaction ecological processes as affectedby land-use change result in a change in ecological conditions G such changes inecological conditions may result in changes in attitudes (even if human perceptionpreviously ignored ecological pattern and process) and changed ecological conditionsare perceived as good or bad by humans H changes in perception and attitude feedback to the societal system (patterns and processes of society) to influence decision-making and this part of the cycle begins anew I in some cases changed ecologicalconditions can alter the coarse-scale environmental context (example urban heatisland) resulting in a feedback that is relatively independent of human response J KWhen a societal response to changed ecological conditions is deemed necessary thesociety can act directly on the changed conditions (J) or on the underlying ecologicalpatterns and processes producing the problem (K) Finally the environmental contextof course influences ecological patterns independent of land use (L)

A modeling framework for investigating citiesThe conceptual insights of the broad scheme in Figure 3illustrate in general how processes from the social realmbiological environment civil infrastructure and the larg-er climatic and geological context can be integrated withthe specific characteristics of metropolitan Phoenix andmetropolitan Baltimore But the extreme patchiness ofurban environments dictates attention to spatial detailusing novel approaches of landscape analysis and model-ing Although the two cities are quite different integra-tion is furthered by a common approach to spatial analy-ses a hierarchical patch dynamics approach (Wu andLoucks 1995 Pickett et al 1999) Hierarchical patchdynamics models start with ecological processes associat-ed with fundamental units of landscapes at some specif-ic scale called patches They then address ecologicalprocesses that are associated with the patches The struc-ture of the patches can be a major determinant of thoseprocesses However patch structure and arrangementalso can change through time Hence models mustaccount for such change that is they must be dynamicFurthermore because patches at a particular scale areoften themselves composed of smaller patches and canbe aggregated into larger patches the models must behierarchical By considering patch dynamics simultane-ously at multiple scales with an accompanying hierarchyof models the complexity of urban systems is rendered

more tractable and translation of information acrossscales is facilitated (Wu 1999)

Special characteristics that dictate a novelframework The modern metropolis presents a striking-ly heterogeneous pattern for study For instance the sharpcontrasts between neighborhoods is a familiar characteris-tic of cities (Clay 1973) Within the span of a city blockwhich is on the order of 200 or fewer meters an observermay cross several obvious boundaries Different kinds ofcommercial use shifts between owner-occupied and rentalproperties and shifts in socioeconomic resources availableto residents are but some of the many contrasts cities pre-sent Such heterogeneity is not unique to dense centralurban districts In fact the contemporary suburb is zonedfor even more discrete transitions than the traditionalmixed use of older cities Residential streets feeder streetscommercial streets strip malls regional malls and indus-trial parks are notable patches in the suburban landscapeOf course the scale of transition in postndashWorld War IIsuburbs tends to be coarser than that of older neighbor-hoods and districts because of the shift to dependence onthe automobile However spatial patchiness in the socialeconomic and infrastructural fabric of metropolitan areasremains their most obvious feature Social scientists havelong been aware of the functional significance of spatial het-erogeneity and mixture of uses within urban areas (Jacobs

578 BioScience July 2000 Vol 50 No 7

Articles

Figure 4 Conceptual scheme adapted to show the interaction among physical ecological engineering social andmanagement variables and drivers in the new Tempe Town Lake Arizona

1962) The ecological significance of such socioeconomicheterogeneity is however an open question Indeed deter-mining to what extent the well-recognized patchiness ofurban areas has ecological dimensions and ecologicalimplications is one of the main motivations of integratedlong-term ecological research in the metropolis

Whatever its ecological significance the conspicuousspatial heterogeneity in urban systems is an entry point forintegration with social science The existence of suchclearly defined patches as neighborhoods and cityscapeswhich combine infrastructural and natural features isapparent to all researchers who must work together togenerate the interdisciplinary synthesis for understandingcities as ecological systems Hydrologists ecologistsdemographers economists engineers and citizens all canand do recognize the spatial heterogeneity of cities Neigh-borhood associations watershed associations censustracts and similar groupings are institutional expressionsof this common recognition Of course each discipline orconstituency may see the boundaries or the most salientfeatures of the patchwork somewhat differently For exam-ple the civil engineer and the urban recreationist will havedifferent views of the boundaries of a watershed The firstmay see a ldquosewershedrdquo the second a visually unified land-scape that is engaging on a morningjog Therefore new multidimensionalclassifications of the heterogeneity ofthe metropolis are required Amongthe principal dimensions of such classi-fications however will be factors thatcontrol the flow of materials energyand information through and withinthe metropolis An emphasis on thesekinds of variables suggests that a spatially explicit ecosys-tem perspective can emerge from the heterogeneity of themetropolis

The interdisciplinary integration required to under-stand the ecological significance of spatial pattern inurban ecosystems must account for several important fea-tures of humans and their institutions Although these fea-tures are implied in the social drivers and phenomena weintroduced earlier (Figure 2 see box page 576) these fea-tures add complexity to ecological studies of the city andtherefore cannot be ignored The spatial heterogeneity ofurban systems is established by formal institutions such aszoning regulations and maintained by other formal insti-tutions such as public works and the courts However lessformal institutions such as families and community asso-ciations also contribute to the spatial structure and itsfunction Humans as individuals and groups are self-aware capable of learning quickly and engaged in exten-sive networks of rapid communication These features ofthe human components of urban systems mean that thefeedback among the biological human infrastructuraland the larger physical contexts can be strong and inmany cases rapid This is one reason that education has

been incorporated into the structure of urban LTER pro-grams We hypothesize that learning about the hetero-geneity and function of an urban area can be a tool thatcitizens and institutions can demonstrably use to improvetheir neighborhoods city and region through manage-ment planning and policy (Grove and Burch 1997)

Patch dynamics and hierarchical approachesSpatial heterogeneity was independently chosen as a start-ing point by both the Central ArizonandashPhoenix and Balti-more Ecosystem Study LTER teams In addition to theadvantages already laid out the patch dynamics approachalso brings the advantage of hierarchical nesting andaggregation Such a flexible hierarchical approach isimportant because the structures and consequently theprocesses that govern the function of the metropolis as anecological system occur at a variety of scales For examplejust because some social processes occur at the scale ofneighborhoods of say 15 square blocks does not meanthat the most important ecological impacts occur at thesame scale Similarly the concentration of resources orwastes in particular patches can be due to decisions madeat great distances from the point of concentration There-fore patches may be scaled up or down for different func-

tional analyses and the configurationof patches that are relevant to specificpaths along which resources and infor-mation flow can be assessed

As an example Baltimore can bedescribed in terms of the five-countymetropolitan area (Figure 5) Within itare three principal watersheds thatextend from the rural hinterlands to the

central city Each of these watersheds can be divided intosubcatchments The 17000-ha Gwynns Falls catchmentcontains 16 smaller watersheds that have been used fordiscussion of management and restoration activities Stillsmaller units can be used for mechanistic studies ofecosystem and socioeconomic processes The fact that dif-ferent nestings are possible within the larger units meansthat the scales at which important interactions occur canbe captured and the promulgation of their effects to dif-ferent scales whether coarser or finer can be determined(see Grove and Burch 1997)

A similar nested set of units is being identified in theCentral ArizonandashPhoenix study area (Figure 5) At thelargest scale three patches exist along an eastndashwest gradi-ent from primarily commercialndashindustrialndashresidential toprimarily agricultural to primarily desert land usendashlandcover The political boundaries of the 24 different munic-ipalities in the Phoenix metropolitan area form anotherset of smaller patches and heterogeneity of land usendashlandcover is evident within these smaller units Interestinglypatch size regularity and connectivity differ within thethree broadest patches (Matthew Luck and Jianguo WuArizona State University personal communication)

July 2000 Vol 50 No 7 BioScience 579

Articles

ldquoThe civil engineer and the

urban recreationist will have

different views of the

boundaries of a watershedrdquo

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

ultimate goal of most inquiries actual investigations maymore often be oriented toward measuring the patterns ofhuman activity these processes create Defining the fol-lowing core topicsmdasheach characterized by its activitiesstructure and historic trajectorymdashis key to a comprehen-sive approach to human ecosystem analysis (Figure 2)

Demographic pattern

Economic system

Power hierarchy

Land use and management

Designed environment

Collectively these core topics serve as guidelines of inquiryanalogous to the five ecological core areas identified earlyin the LTER programrsquos history On the one hand theyreflect the processes operating within the system on theother they are a practical guide for field investigationsThe entire LTER network (not just the urban and aug-mented sites) thus provides an excellent starting point toincorporate social science research that is relevant to andintegrated with studies of ecosystem change

A conceptual scheme for understanding urbanecological systems We have now identified drivers andpatterns of activity in both ecological and social systemsthat must collectively be considered for a full understand-ing of human-dominated ecosystems A more specificconceptual scheme or model for how a study of urban eco-logical systems can be approached is represented by Figure3 This scheme has been modified through recent discus-sions from schemes presented by the Baltimore and

Phoenix LTER groups in grant proposals and various pub-lications (eg Grimm 1997 Pickett et al 1997) indeed itis still evolving The diagram includes a set of variablesthat are linked by interactions and feedbacks

The two variables in the upper cornersmdashcoarse-scaleenvironmental context and societal patterns and process-esmdashcan be viewed as constraints which are the outcomeof operation of the fundamental biogeophysical and soci-etal drivers The coarser-scale environmental contextincludes climate geology history and biogeographicalsetting and societal patterns and processes encompass thesocioeconomic system culture demography and socialinstitutions

The middle two variables are the focus of the new urbanLTER research land use and ecological patterns andprocesses associated with any given land use Land useincorporates both the intent and the reality of how a giv-en parcel within a metropolitan boundary is altered byhuman decisions whereas ecological patterns and process-esmdashenergy flow nutrient cycling the hydrologic cyclespecies distribution abundance and interactions ecosys-tem and landscape structure and disturbancemdashare thosephenomena studied by all LTER teams Land use and asso-ciated ecological conditions are viewed at a single point intime using a hierarchical patch mosaic perspective Land-use change occurs because of development urban renew-al changes in land management or ownership or infra-structure development among other causes

The last two variables in the lower corners of Figure 3mdashchanges in ecological conditions and changes in humanperceptions and attitudesmdashresult from the interactionbetween land-use change and ecological pattern andprocess Changes in ecological conditions represent thenext time step of ecological patterns and processes andchanges in human perceptions and attitudes represent thehuman reaction to either ecological pattern and process orchanges therein as expressed through the filter of humanexperience

The interactions and feedbacks depicted in Figure 3 areintended to reflect temporal dynamics to some extent Asan example land-use decisions are based on the environ-mental and politicondashsocioeconomic context the land useis then perceived as good or bad and this feedback canlead to different human decisions This sequence of deci-sions and changes in land use does not incorporate newecological information (ie the changes in ecological con-dition that result from the land-use change) such a situa-tion is unstable and seems unlikely to lead to a sustainableurban environment A sequence of interactions and feed-backs carried out when a change occurs or in response toan environmental problem however would incorporateeither short-term solutions to those problems or adjust-ments in management decisions based on a solid ecologi-cal foundation

To illustrate the sequence of changes and feedbacks tofurther change consider an example from the Central

576 BioScience July 2000 Vol 50 No 7

Articles

Social institutions Health Justice Faith Commerce Education Leisure Government

Social dynamics Physiological Individual Organizational Institutional Environmental

Social system componentsThe following are examples of social system componentsto be incorporated into humanndashecological models ofurban ecosystems (from Pickett et al 1997)

Social order Age Gender Class Norms Wealth Power Status Knowledge Territory

Social resources Economic (information

population labor) Cultural (organizations

beliefs myths)

ArizonandashPhoenix LTER sitethe establishment of an artifi-cial lake in the once-dry SaltRiver bed (Figure 4) The ini-tial land-use decision (estab-lishment of the lake) was con-strained on the physicalndashecological side by the existenceof an alluvial channel with nosurface water flow (because ofupstream impoundments) in aregion of North America char-acterized by a high propensityfor flash flooding (Baker1977) Societal constraintsincluded the economic cost ofthe project the perceptions ofpolitical and economic bene-fit available technology (col-lapsible dams recirculatingpumps) and the existence ofhuman-created infrastructure(engineered channel diversion ofsurface flow into canals) Giventhese constraints and based onour best understanding of lakeecology the ecological conditionsassociated with the lake when itwas filled were likely to be highnutrients with concomitant highalgal production high rates ofinfiltration and a high probabilityof floods At the next time step weexpected such changes in ecologi-cal conditions as eutrophicationlosses of water to the groundwatersystem and establishment of arobust mosquito population Pre-liminary monitoring confirms thepredictions of high phytoplank-ton biomass and insect popula-tions and a summer flash floodresulted in a brief episode of high phosphorus loading(Amalfi 1999)

The predicted sequence of ecological conditions mayresult in a full range of reactions from the public Feed-backs to the societal patterns and processesmdashin this casethe design and management decisionsmdashcould lead eitherto chemical additions to control algal blooms or to moreecologically based management practices such as controlof nutrient inputs Indeed complaints from the publicabout insect populations already have resulted in imple-mentation of a chemical control program (Amalfi 1999)and feedbacks of ecological information and knowledge ofthe lakersquos dynamics have helped create and modify a lakemanagement plan Thus societal responses can act directly

on the changed conditions (arrow J in Figure 3 eg addi-tion of chemical control agents the technological solutionin Figure 4) or indirectly via an effect on the underlyingecological patterns and processes producing the problem(arrow K in Figure 3 eg diversion of upstream point-source nutrient inputs to the water supply the arrowlabeled management for sustainability in Figure 4) Webelieve that this general scheme (Figure 3) allows us tothink about how social and ecological systems interact inurban areas In addition to addressing fundamental eco-logical processes this approach allows one to make pre-dictions that are testable and it also provides tools that cit-izens and decision-makers can use to create moreecologically sound policy

July 2000 Vol 50 No 7 BioScience 577

Articles

Figure 3 Conceptual scheme for integrating ecological and social systems in urbanenvironments Variables are in boxes interactions and feedbacks are arrows Aenvironmental context sets the range of possibilities for land usendashland cover B societaldecisions and human behavior (incorporating their suite of determinants) are thedirect drivers of land-use change C the pattern of land use (whatever the driver)determines ecological patterns and processes D humans perceive and react to land-usechange (independent of any ecological effects) E humans also perceive and react toecological patterns and processes F in this interaction ecological processes as affectedby land-use change result in a change in ecological conditions G such changes inecological conditions may result in changes in attitudes (even if human perceptionpreviously ignored ecological pattern and process) and changed ecological conditionsare perceived as good or bad by humans H changes in perception and attitude feedback to the societal system (patterns and processes of society) to influence decision-making and this part of the cycle begins anew I in some cases changed ecologicalconditions can alter the coarse-scale environmental context (example urban heatisland) resulting in a feedback that is relatively independent of human response J KWhen a societal response to changed ecological conditions is deemed necessary thesociety can act directly on the changed conditions (J) or on the underlying ecologicalpatterns and processes producing the problem (K) Finally the environmental contextof course influences ecological patterns independent of land use (L)

A modeling framework for investigating citiesThe conceptual insights of the broad scheme in Figure 3illustrate in general how processes from the social realmbiological environment civil infrastructure and the larg-er climatic and geological context can be integrated withthe specific characteristics of metropolitan Phoenix andmetropolitan Baltimore But the extreme patchiness ofurban environments dictates attention to spatial detailusing novel approaches of landscape analysis and model-ing Although the two cities are quite different integra-tion is furthered by a common approach to spatial analy-ses a hierarchical patch dynamics approach (Wu andLoucks 1995 Pickett et al 1999) Hierarchical patchdynamics models start with ecological processes associat-ed with fundamental units of landscapes at some specif-ic scale called patches They then address ecologicalprocesses that are associated with the patches The struc-ture of the patches can be a major determinant of thoseprocesses However patch structure and arrangementalso can change through time Hence models mustaccount for such change that is they must be dynamicFurthermore because patches at a particular scale areoften themselves composed of smaller patches and canbe aggregated into larger patches the models must behierarchical By considering patch dynamics simultane-ously at multiple scales with an accompanying hierarchyof models the complexity of urban systems is rendered

more tractable and translation of information acrossscales is facilitated (Wu 1999)

Special characteristics that dictate a novelframework The modern metropolis presents a striking-ly heterogeneous pattern for study For instance the sharpcontrasts between neighborhoods is a familiar characteris-tic of cities (Clay 1973) Within the span of a city blockwhich is on the order of 200 or fewer meters an observermay cross several obvious boundaries Different kinds ofcommercial use shifts between owner-occupied and rentalproperties and shifts in socioeconomic resources availableto residents are but some of the many contrasts cities pre-sent Such heterogeneity is not unique to dense centralurban districts In fact the contemporary suburb is zonedfor even more discrete transitions than the traditionalmixed use of older cities Residential streets feeder streetscommercial streets strip malls regional malls and indus-trial parks are notable patches in the suburban landscapeOf course the scale of transition in postndashWorld War IIsuburbs tends to be coarser than that of older neighbor-hoods and districts because of the shift to dependence onthe automobile However spatial patchiness in the socialeconomic and infrastructural fabric of metropolitan areasremains their most obvious feature Social scientists havelong been aware of the functional significance of spatial het-erogeneity and mixture of uses within urban areas (Jacobs

578 BioScience July 2000 Vol 50 No 7

Articles

Figure 4 Conceptual scheme adapted to show the interaction among physical ecological engineering social andmanagement variables and drivers in the new Tempe Town Lake Arizona

1962) The ecological significance of such socioeconomicheterogeneity is however an open question Indeed deter-mining to what extent the well-recognized patchiness ofurban areas has ecological dimensions and ecologicalimplications is one of the main motivations of integratedlong-term ecological research in the metropolis

Whatever its ecological significance the conspicuousspatial heterogeneity in urban systems is an entry point forintegration with social science The existence of suchclearly defined patches as neighborhoods and cityscapeswhich combine infrastructural and natural features isapparent to all researchers who must work together togenerate the interdisciplinary synthesis for understandingcities as ecological systems Hydrologists ecologistsdemographers economists engineers and citizens all canand do recognize the spatial heterogeneity of cities Neigh-borhood associations watershed associations censustracts and similar groupings are institutional expressionsof this common recognition Of course each discipline orconstituency may see the boundaries or the most salientfeatures of the patchwork somewhat differently For exam-ple the civil engineer and the urban recreationist will havedifferent views of the boundaries of a watershed The firstmay see a ldquosewershedrdquo the second a visually unified land-scape that is engaging on a morningjog Therefore new multidimensionalclassifications of the heterogeneity ofthe metropolis are required Amongthe principal dimensions of such classi-fications however will be factors thatcontrol the flow of materials energyand information through and withinthe metropolis An emphasis on thesekinds of variables suggests that a spatially explicit ecosys-tem perspective can emerge from the heterogeneity of themetropolis

The interdisciplinary integration required to under-stand the ecological significance of spatial pattern inurban ecosystems must account for several important fea-tures of humans and their institutions Although these fea-tures are implied in the social drivers and phenomena weintroduced earlier (Figure 2 see box page 576) these fea-tures add complexity to ecological studies of the city andtherefore cannot be ignored The spatial heterogeneity ofurban systems is established by formal institutions such aszoning regulations and maintained by other formal insti-tutions such as public works and the courts However lessformal institutions such as families and community asso-ciations also contribute to the spatial structure and itsfunction Humans as individuals and groups are self-aware capable of learning quickly and engaged in exten-sive networks of rapid communication These features ofthe human components of urban systems mean that thefeedback among the biological human infrastructuraland the larger physical contexts can be strong and inmany cases rapid This is one reason that education has

been incorporated into the structure of urban LTER pro-grams We hypothesize that learning about the hetero-geneity and function of an urban area can be a tool thatcitizens and institutions can demonstrably use to improvetheir neighborhoods city and region through manage-ment planning and policy (Grove and Burch 1997)

Patch dynamics and hierarchical approachesSpatial heterogeneity was independently chosen as a start-ing point by both the Central ArizonandashPhoenix and Balti-more Ecosystem Study LTER teams In addition to theadvantages already laid out the patch dynamics approachalso brings the advantage of hierarchical nesting andaggregation Such a flexible hierarchical approach isimportant because the structures and consequently theprocesses that govern the function of the metropolis as anecological system occur at a variety of scales For examplejust because some social processes occur at the scale ofneighborhoods of say 15 square blocks does not meanthat the most important ecological impacts occur at thesame scale Similarly the concentration of resources orwastes in particular patches can be due to decisions madeat great distances from the point of concentration There-fore patches may be scaled up or down for different func-

tional analyses and the configurationof patches that are relevant to specificpaths along which resources and infor-mation flow can be assessed

As an example Baltimore can bedescribed in terms of the five-countymetropolitan area (Figure 5) Within itare three principal watersheds thatextend from the rural hinterlands to the

central city Each of these watersheds can be divided intosubcatchments The 17000-ha Gwynns Falls catchmentcontains 16 smaller watersheds that have been used fordiscussion of management and restoration activities Stillsmaller units can be used for mechanistic studies ofecosystem and socioeconomic processes The fact that dif-ferent nestings are possible within the larger units meansthat the scales at which important interactions occur canbe captured and the promulgation of their effects to dif-ferent scales whether coarser or finer can be determined(see Grove and Burch 1997)

A similar nested set of units is being identified in theCentral ArizonandashPhoenix study area (Figure 5) At thelargest scale three patches exist along an eastndashwest gradi-ent from primarily commercialndashindustrialndashresidential toprimarily agricultural to primarily desert land usendashlandcover The political boundaries of the 24 different munic-ipalities in the Phoenix metropolitan area form anotherset of smaller patches and heterogeneity of land usendashlandcover is evident within these smaller units Interestinglypatch size regularity and connectivity differ within thethree broadest patches (Matthew Luck and Jianguo WuArizona State University personal communication)

July 2000 Vol 50 No 7 BioScience 579

Articles

ldquoThe civil engineer and the

urban recreationist will have

different views of the

boundaries of a watershedrdquo

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

ArizonandashPhoenix LTER sitethe establishment of an artifi-cial lake in the once-dry SaltRiver bed (Figure 4) The ini-tial land-use decision (estab-lishment of the lake) was con-strained on the physicalndashecological side by the existenceof an alluvial channel with nosurface water flow (because ofupstream impoundments) in aregion of North America char-acterized by a high propensityfor flash flooding (Baker1977) Societal constraintsincluded the economic cost ofthe project the perceptions ofpolitical and economic bene-fit available technology (col-lapsible dams recirculatingpumps) and the existence ofhuman-created infrastructure(engineered channel diversion ofsurface flow into canals) Giventhese constraints and based onour best understanding of lakeecology the ecological conditionsassociated with the lake when itwas filled were likely to be highnutrients with concomitant highalgal production high rates ofinfiltration and a high probabilityof floods At the next time step weexpected such changes in ecologi-cal conditions as eutrophicationlosses of water to the groundwatersystem and establishment of arobust mosquito population Pre-liminary monitoring confirms thepredictions of high phytoplank-ton biomass and insect popula-tions and a summer flash floodresulted in a brief episode of high phosphorus loading(Amalfi 1999)

The predicted sequence of ecological conditions mayresult in a full range of reactions from the public Feed-backs to the societal patterns and processesmdashin this casethe design and management decisionsmdashcould lead eitherto chemical additions to control algal blooms or to moreecologically based management practices such as controlof nutrient inputs Indeed complaints from the publicabout insect populations already have resulted in imple-mentation of a chemical control program (Amalfi 1999)and feedbacks of ecological information and knowledge ofthe lakersquos dynamics have helped create and modify a lakemanagement plan Thus societal responses can act directly

on the changed conditions (arrow J in Figure 3 eg addi-tion of chemical control agents the technological solutionin Figure 4) or indirectly via an effect on the underlyingecological patterns and processes producing the problem(arrow K in Figure 3 eg diversion of upstream point-source nutrient inputs to the water supply the arrowlabeled management for sustainability in Figure 4) Webelieve that this general scheme (Figure 3) allows us tothink about how social and ecological systems interact inurban areas In addition to addressing fundamental eco-logical processes this approach allows one to make pre-dictions that are testable and it also provides tools that cit-izens and decision-makers can use to create moreecologically sound policy

July 2000 Vol 50 No 7 BioScience 577

Articles

Figure 3 Conceptual scheme for integrating ecological and social systems in urbanenvironments Variables are in boxes interactions and feedbacks are arrows Aenvironmental context sets the range of possibilities for land usendashland cover B societaldecisions and human behavior (incorporating their suite of determinants) are thedirect drivers of land-use change C the pattern of land use (whatever the driver)determines ecological patterns and processes D humans perceive and react to land-usechange (independent of any ecological effects) E humans also perceive and react toecological patterns and processes F in this interaction ecological processes as affectedby land-use change result in a change in ecological conditions G such changes inecological conditions may result in changes in attitudes (even if human perceptionpreviously ignored ecological pattern and process) and changed ecological conditionsare perceived as good or bad by humans H changes in perception and attitude feedback to the societal system (patterns and processes of society) to influence decision-making and this part of the cycle begins anew I in some cases changed ecologicalconditions can alter the coarse-scale environmental context (example urban heatisland) resulting in a feedback that is relatively independent of human response J KWhen a societal response to changed ecological conditions is deemed necessary thesociety can act directly on the changed conditions (J) or on the underlying ecologicalpatterns and processes producing the problem (K) Finally the environmental contextof course influences ecological patterns independent of land use (L)

A modeling framework for investigating citiesThe conceptual insights of the broad scheme in Figure 3illustrate in general how processes from the social realmbiological environment civil infrastructure and the larg-er climatic and geological context can be integrated withthe specific characteristics of metropolitan Phoenix andmetropolitan Baltimore But the extreme patchiness ofurban environments dictates attention to spatial detailusing novel approaches of landscape analysis and model-ing Although the two cities are quite different integra-tion is furthered by a common approach to spatial analy-ses a hierarchical patch dynamics approach (Wu andLoucks 1995 Pickett et al 1999) Hierarchical patchdynamics models start with ecological processes associat-ed with fundamental units of landscapes at some specif-ic scale called patches They then address ecologicalprocesses that are associated with the patches The struc-ture of the patches can be a major determinant of thoseprocesses However patch structure and arrangementalso can change through time Hence models mustaccount for such change that is they must be dynamicFurthermore because patches at a particular scale areoften themselves composed of smaller patches and canbe aggregated into larger patches the models must behierarchical By considering patch dynamics simultane-ously at multiple scales with an accompanying hierarchyof models the complexity of urban systems is rendered

more tractable and translation of information acrossscales is facilitated (Wu 1999)

Special characteristics that dictate a novelframework The modern metropolis presents a striking-ly heterogeneous pattern for study For instance the sharpcontrasts between neighborhoods is a familiar characteris-tic of cities (Clay 1973) Within the span of a city blockwhich is on the order of 200 or fewer meters an observermay cross several obvious boundaries Different kinds ofcommercial use shifts between owner-occupied and rentalproperties and shifts in socioeconomic resources availableto residents are but some of the many contrasts cities pre-sent Such heterogeneity is not unique to dense centralurban districts In fact the contemporary suburb is zonedfor even more discrete transitions than the traditionalmixed use of older cities Residential streets feeder streetscommercial streets strip malls regional malls and indus-trial parks are notable patches in the suburban landscapeOf course the scale of transition in postndashWorld War IIsuburbs tends to be coarser than that of older neighbor-hoods and districts because of the shift to dependence onthe automobile However spatial patchiness in the socialeconomic and infrastructural fabric of metropolitan areasremains their most obvious feature Social scientists havelong been aware of the functional significance of spatial het-erogeneity and mixture of uses within urban areas (Jacobs

578 BioScience July 2000 Vol 50 No 7

Articles

Figure 4 Conceptual scheme adapted to show the interaction among physical ecological engineering social andmanagement variables and drivers in the new Tempe Town Lake Arizona

1962) The ecological significance of such socioeconomicheterogeneity is however an open question Indeed deter-mining to what extent the well-recognized patchiness ofurban areas has ecological dimensions and ecologicalimplications is one of the main motivations of integratedlong-term ecological research in the metropolis

Whatever its ecological significance the conspicuousspatial heterogeneity in urban systems is an entry point forintegration with social science The existence of suchclearly defined patches as neighborhoods and cityscapeswhich combine infrastructural and natural features isapparent to all researchers who must work together togenerate the interdisciplinary synthesis for understandingcities as ecological systems Hydrologists ecologistsdemographers economists engineers and citizens all canand do recognize the spatial heterogeneity of cities Neigh-borhood associations watershed associations censustracts and similar groupings are institutional expressionsof this common recognition Of course each discipline orconstituency may see the boundaries or the most salientfeatures of the patchwork somewhat differently For exam-ple the civil engineer and the urban recreationist will havedifferent views of the boundaries of a watershed The firstmay see a ldquosewershedrdquo the second a visually unified land-scape that is engaging on a morningjog Therefore new multidimensionalclassifications of the heterogeneity ofthe metropolis are required Amongthe principal dimensions of such classi-fications however will be factors thatcontrol the flow of materials energyand information through and withinthe metropolis An emphasis on thesekinds of variables suggests that a spatially explicit ecosys-tem perspective can emerge from the heterogeneity of themetropolis

The interdisciplinary integration required to under-stand the ecological significance of spatial pattern inurban ecosystems must account for several important fea-tures of humans and their institutions Although these fea-tures are implied in the social drivers and phenomena weintroduced earlier (Figure 2 see box page 576) these fea-tures add complexity to ecological studies of the city andtherefore cannot be ignored The spatial heterogeneity ofurban systems is established by formal institutions such aszoning regulations and maintained by other formal insti-tutions such as public works and the courts However lessformal institutions such as families and community asso-ciations also contribute to the spatial structure and itsfunction Humans as individuals and groups are self-aware capable of learning quickly and engaged in exten-sive networks of rapid communication These features ofthe human components of urban systems mean that thefeedback among the biological human infrastructuraland the larger physical contexts can be strong and inmany cases rapid This is one reason that education has

been incorporated into the structure of urban LTER pro-grams We hypothesize that learning about the hetero-geneity and function of an urban area can be a tool thatcitizens and institutions can demonstrably use to improvetheir neighborhoods city and region through manage-ment planning and policy (Grove and Burch 1997)

Patch dynamics and hierarchical approachesSpatial heterogeneity was independently chosen as a start-ing point by both the Central ArizonandashPhoenix and Balti-more Ecosystem Study LTER teams In addition to theadvantages already laid out the patch dynamics approachalso brings the advantage of hierarchical nesting andaggregation Such a flexible hierarchical approach isimportant because the structures and consequently theprocesses that govern the function of the metropolis as anecological system occur at a variety of scales For examplejust because some social processes occur at the scale ofneighborhoods of say 15 square blocks does not meanthat the most important ecological impacts occur at thesame scale Similarly the concentration of resources orwastes in particular patches can be due to decisions madeat great distances from the point of concentration There-fore patches may be scaled up or down for different func-

tional analyses and the configurationof patches that are relevant to specificpaths along which resources and infor-mation flow can be assessed

As an example Baltimore can bedescribed in terms of the five-countymetropolitan area (Figure 5) Within itare three principal watersheds thatextend from the rural hinterlands to the

central city Each of these watersheds can be divided intosubcatchments The 17000-ha Gwynns Falls catchmentcontains 16 smaller watersheds that have been used fordiscussion of management and restoration activities Stillsmaller units can be used for mechanistic studies ofecosystem and socioeconomic processes The fact that dif-ferent nestings are possible within the larger units meansthat the scales at which important interactions occur canbe captured and the promulgation of their effects to dif-ferent scales whether coarser or finer can be determined(see Grove and Burch 1997)

A similar nested set of units is being identified in theCentral ArizonandashPhoenix study area (Figure 5) At thelargest scale three patches exist along an eastndashwest gradi-ent from primarily commercialndashindustrialndashresidential toprimarily agricultural to primarily desert land usendashlandcover The political boundaries of the 24 different munic-ipalities in the Phoenix metropolitan area form anotherset of smaller patches and heterogeneity of land usendashlandcover is evident within these smaller units Interestinglypatch size regularity and connectivity differ within thethree broadest patches (Matthew Luck and Jianguo WuArizona State University personal communication)

July 2000 Vol 50 No 7 BioScience 579

Articles

ldquoThe civil engineer and the

urban recreationist will have

different views of the

boundaries of a watershedrdquo

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

A modeling framework for investigating citiesThe conceptual insights of the broad scheme in Figure 3illustrate in general how processes from the social realmbiological environment civil infrastructure and the larg-er climatic and geological context can be integrated withthe specific characteristics of metropolitan Phoenix andmetropolitan Baltimore But the extreme patchiness ofurban environments dictates attention to spatial detailusing novel approaches of landscape analysis and model-ing Although the two cities are quite different integra-tion is furthered by a common approach to spatial analy-ses a hierarchical patch dynamics approach (Wu andLoucks 1995 Pickett et al 1999) Hierarchical patchdynamics models start with ecological processes associat-ed with fundamental units of landscapes at some specif-ic scale called patches They then address ecologicalprocesses that are associated with the patches The struc-ture of the patches can be a major determinant of thoseprocesses However patch structure and arrangementalso can change through time Hence models mustaccount for such change that is they must be dynamicFurthermore because patches at a particular scale areoften themselves composed of smaller patches and canbe aggregated into larger patches the models must behierarchical By considering patch dynamics simultane-ously at multiple scales with an accompanying hierarchyof models the complexity of urban systems is rendered

more tractable and translation of information acrossscales is facilitated (Wu 1999)

Special characteristics that dictate a novelframework The modern metropolis presents a striking-ly heterogeneous pattern for study For instance the sharpcontrasts between neighborhoods is a familiar characteris-tic of cities (Clay 1973) Within the span of a city blockwhich is on the order of 200 or fewer meters an observermay cross several obvious boundaries Different kinds ofcommercial use shifts between owner-occupied and rentalproperties and shifts in socioeconomic resources availableto residents are but some of the many contrasts cities pre-sent Such heterogeneity is not unique to dense centralurban districts In fact the contemporary suburb is zonedfor even more discrete transitions than the traditionalmixed use of older cities Residential streets feeder streetscommercial streets strip malls regional malls and indus-trial parks are notable patches in the suburban landscapeOf course the scale of transition in postndashWorld War IIsuburbs tends to be coarser than that of older neighbor-hoods and districts because of the shift to dependence onthe automobile However spatial patchiness in the socialeconomic and infrastructural fabric of metropolitan areasremains their most obvious feature Social scientists havelong been aware of the functional significance of spatial het-erogeneity and mixture of uses within urban areas (Jacobs

578 BioScience July 2000 Vol 50 No 7

Articles

Figure 4 Conceptual scheme adapted to show the interaction among physical ecological engineering social andmanagement variables and drivers in the new Tempe Town Lake Arizona

1962) The ecological significance of such socioeconomicheterogeneity is however an open question Indeed deter-mining to what extent the well-recognized patchiness ofurban areas has ecological dimensions and ecologicalimplications is one of the main motivations of integratedlong-term ecological research in the metropolis

Whatever its ecological significance the conspicuousspatial heterogeneity in urban systems is an entry point forintegration with social science The existence of suchclearly defined patches as neighborhoods and cityscapeswhich combine infrastructural and natural features isapparent to all researchers who must work together togenerate the interdisciplinary synthesis for understandingcities as ecological systems Hydrologists ecologistsdemographers economists engineers and citizens all canand do recognize the spatial heterogeneity of cities Neigh-borhood associations watershed associations censustracts and similar groupings are institutional expressionsof this common recognition Of course each discipline orconstituency may see the boundaries or the most salientfeatures of the patchwork somewhat differently For exam-ple the civil engineer and the urban recreationist will havedifferent views of the boundaries of a watershed The firstmay see a ldquosewershedrdquo the second a visually unified land-scape that is engaging on a morningjog Therefore new multidimensionalclassifications of the heterogeneity ofthe metropolis are required Amongthe principal dimensions of such classi-fications however will be factors thatcontrol the flow of materials energyand information through and withinthe metropolis An emphasis on thesekinds of variables suggests that a spatially explicit ecosys-tem perspective can emerge from the heterogeneity of themetropolis

The interdisciplinary integration required to under-stand the ecological significance of spatial pattern inurban ecosystems must account for several important fea-tures of humans and their institutions Although these fea-tures are implied in the social drivers and phenomena weintroduced earlier (Figure 2 see box page 576) these fea-tures add complexity to ecological studies of the city andtherefore cannot be ignored The spatial heterogeneity ofurban systems is established by formal institutions such aszoning regulations and maintained by other formal insti-tutions such as public works and the courts However lessformal institutions such as families and community asso-ciations also contribute to the spatial structure and itsfunction Humans as individuals and groups are self-aware capable of learning quickly and engaged in exten-sive networks of rapid communication These features ofthe human components of urban systems mean that thefeedback among the biological human infrastructuraland the larger physical contexts can be strong and inmany cases rapid This is one reason that education has

been incorporated into the structure of urban LTER pro-grams We hypothesize that learning about the hetero-geneity and function of an urban area can be a tool thatcitizens and institutions can demonstrably use to improvetheir neighborhoods city and region through manage-ment planning and policy (Grove and Burch 1997)

Patch dynamics and hierarchical approachesSpatial heterogeneity was independently chosen as a start-ing point by both the Central ArizonandashPhoenix and Balti-more Ecosystem Study LTER teams In addition to theadvantages already laid out the patch dynamics approachalso brings the advantage of hierarchical nesting andaggregation Such a flexible hierarchical approach isimportant because the structures and consequently theprocesses that govern the function of the metropolis as anecological system occur at a variety of scales For examplejust because some social processes occur at the scale ofneighborhoods of say 15 square blocks does not meanthat the most important ecological impacts occur at thesame scale Similarly the concentration of resources orwastes in particular patches can be due to decisions madeat great distances from the point of concentration There-fore patches may be scaled up or down for different func-

tional analyses and the configurationof patches that are relevant to specificpaths along which resources and infor-mation flow can be assessed

As an example Baltimore can bedescribed in terms of the five-countymetropolitan area (Figure 5) Within itare three principal watersheds thatextend from the rural hinterlands to the

central city Each of these watersheds can be divided intosubcatchments The 17000-ha Gwynns Falls catchmentcontains 16 smaller watersheds that have been used fordiscussion of management and restoration activities Stillsmaller units can be used for mechanistic studies ofecosystem and socioeconomic processes The fact that dif-ferent nestings are possible within the larger units meansthat the scales at which important interactions occur canbe captured and the promulgation of their effects to dif-ferent scales whether coarser or finer can be determined(see Grove and Burch 1997)

A similar nested set of units is being identified in theCentral ArizonandashPhoenix study area (Figure 5) At thelargest scale three patches exist along an eastndashwest gradi-ent from primarily commercialndashindustrialndashresidential toprimarily agricultural to primarily desert land usendashlandcover The political boundaries of the 24 different munic-ipalities in the Phoenix metropolitan area form anotherset of smaller patches and heterogeneity of land usendashlandcover is evident within these smaller units Interestinglypatch size regularity and connectivity differ within thethree broadest patches (Matthew Luck and Jianguo WuArizona State University personal communication)

July 2000 Vol 50 No 7 BioScience 579

Articles

ldquoThe civil engineer and the

urban recreationist will have

different views of the

boundaries of a watershedrdquo

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

1962) The ecological significance of such socioeconomicheterogeneity is however an open question Indeed deter-mining to what extent the well-recognized patchiness ofurban areas has ecological dimensions and ecologicalimplications is one of the main motivations of integratedlong-term ecological research in the metropolis

Whatever its ecological significance the conspicuousspatial heterogeneity in urban systems is an entry point forintegration with social science The existence of suchclearly defined patches as neighborhoods and cityscapeswhich combine infrastructural and natural features isapparent to all researchers who must work together togenerate the interdisciplinary synthesis for understandingcities as ecological systems Hydrologists ecologistsdemographers economists engineers and citizens all canand do recognize the spatial heterogeneity of cities Neigh-borhood associations watershed associations censustracts and similar groupings are institutional expressionsof this common recognition Of course each discipline orconstituency may see the boundaries or the most salientfeatures of the patchwork somewhat differently For exam-ple the civil engineer and the urban recreationist will havedifferent views of the boundaries of a watershed The firstmay see a ldquosewershedrdquo the second a visually unified land-scape that is engaging on a morningjog Therefore new multidimensionalclassifications of the heterogeneity ofthe metropolis are required Amongthe principal dimensions of such classi-fications however will be factors thatcontrol the flow of materials energyand information through and withinthe metropolis An emphasis on thesekinds of variables suggests that a spatially explicit ecosys-tem perspective can emerge from the heterogeneity of themetropolis

The interdisciplinary integration required to under-stand the ecological significance of spatial pattern inurban ecosystems must account for several important fea-tures of humans and their institutions Although these fea-tures are implied in the social drivers and phenomena weintroduced earlier (Figure 2 see box page 576) these fea-tures add complexity to ecological studies of the city andtherefore cannot be ignored The spatial heterogeneity ofurban systems is established by formal institutions such aszoning regulations and maintained by other formal insti-tutions such as public works and the courts However lessformal institutions such as families and community asso-ciations also contribute to the spatial structure and itsfunction Humans as individuals and groups are self-aware capable of learning quickly and engaged in exten-sive networks of rapid communication These features ofthe human components of urban systems mean that thefeedback among the biological human infrastructuraland the larger physical contexts can be strong and inmany cases rapid This is one reason that education has

been incorporated into the structure of urban LTER pro-grams We hypothesize that learning about the hetero-geneity and function of an urban area can be a tool thatcitizens and institutions can demonstrably use to improvetheir neighborhoods city and region through manage-ment planning and policy (Grove and Burch 1997)

Patch dynamics and hierarchical approachesSpatial heterogeneity was independently chosen as a start-ing point by both the Central ArizonandashPhoenix and Balti-more Ecosystem Study LTER teams In addition to theadvantages already laid out the patch dynamics approachalso brings the advantage of hierarchical nesting andaggregation Such a flexible hierarchical approach isimportant because the structures and consequently theprocesses that govern the function of the metropolis as anecological system occur at a variety of scales For examplejust because some social processes occur at the scale ofneighborhoods of say 15 square blocks does not meanthat the most important ecological impacts occur at thesame scale Similarly the concentration of resources orwastes in particular patches can be due to decisions madeat great distances from the point of concentration There-fore patches may be scaled up or down for different func-

tional analyses and the configurationof patches that are relevant to specificpaths along which resources and infor-mation flow can be assessed

As an example Baltimore can bedescribed in terms of the five-countymetropolitan area (Figure 5) Within itare three principal watersheds thatextend from the rural hinterlands to the

central city Each of these watersheds can be divided intosubcatchments The 17000-ha Gwynns Falls catchmentcontains 16 smaller watersheds that have been used fordiscussion of management and restoration activities Stillsmaller units can be used for mechanistic studies ofecosystem and socioeconomic processes The fact that dif-ferent nestings are possible within the larger units meansthat the scales at which important interactions occur canbe captured and the promulgation of their effects to dif-ferent scales whether coarser or finer can be determined(see Grove and Burch 1997)

A similar nested set of units is being identified in theCentral ArizonandashPhoenix study area (Figure 5) At thelargest scale three patches exist along an eastndashwest gradi-ent from primarily commercialndashindustrialndashresidential toprimarily agricultural to primarily desert land usendashlandcover The political boundaries of the 24 different munic-ipalities in the Phoenix metropolitan area form anotherset of smaller patches and heterogeneity of land usendashlandcover is evident within these smaller units Interestinglypatch size regularity and connectivity differ within thethree broadest patches (Matthew Luck and Jianguo WuArizona State University personal communication)

July 2000 Vol 50 No 7 BioScience 579

Articles

ldquoThe civil engineer and the

urban recreationist will have

different views of the

boundaries of a watershedrdquo

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

The patch dynamics approach focuses explicitly notonly on the spatial pattern of heterogeneity at a given timebut also on how and why the pattern changes throughtime and on how that pattern affects ecological and socialprocesses Because cities are both expanding and changingwithin their boundaries the dynamic aspect of this ap-proach is crucial to complete understanding of urban eco-logical systems Even within a coarse resolution land-

cover type change occurs over time in the resources avail-able for management of biotic structure and maintenanceof civil infrastructure and in the requirements and inter-ests of humans in specific patches The explosive changesin patch structure at the urban fringe of the rapidlyexpanding Phoenix metropolis is one example of this phe-nomenon formerly desert patches are converted to resi-dential housing developments and with this conversion

580 BioScience July 2000 Vol 50 No 7

Articles

Figure 5 Examples of hierarchically nested patch structure at three scales in the Central ArizonandashPhoenix (CAP upper panels)and Baltimore Ecosystem Study (BES lower panels) regions At the broadest scale (A D) patches in the CAP study area includedesert (mustard) agriculture (green) and urban (blue) for the BES patches are rural (green) urban (yellow) and aquatic(blue) (B) The municipality of Scottsdale AZ showing major areas of urbanndashresidential development (blue lower portion)and undeveloped open lands (tan developable brown dedicated) (C) Enlargement of rectangle in B showing additional patchstructure at a neighborhood scale (green golf coursepark mustard undeveloped desert red vacant pink xeric residentialpurple mesic residential yellow asphalt) (E) Gwynnrsquos Falls watershed MD with residential (yellow) commercialindustrial(red) agricultural (light green) institutional (medium green) and forest (dark green) patch types (F) Enlargement of rectanglein E showing additional patch structure at a neighborhood scale (dark green pervious surface canopy cover light greenpervious surfaceno canopy cover yellow impervious surfacecanopy cover red impervious surfaceno canopy cover blueneighborhood boundaries black circles abandoned lots) Panel A courtesy of CAP Historic Land Use Project (caplterasueduresearchcontributions HistoricLandUse_Colorpdf) B and C courtesy of CAP LTERGeologic Remote Sensing Laboratory(elwoodlaasuedugrsl) D E and F courtesy of USDA Forest Service and BES LTER

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

demand for amenities of urban life increases sharply InBaltimore patch dynamics are conspicuous both at thesuburban fringe and in the ever-growing collection ofvacant buildings and empty lots within the older denseurban areas In both the fringe and core patches ecologi-cal processing of water nutrients and the provision ofgoods and services are not constant in time The dynamicsof patches in and around the city have implications for theecological processes and status of areas well beyond thecity and even beyond the present suburban and exurbanareas The search for open land development opportuni-ties and changes in the economics of farming and pro-duction forestry all influence and are influenced by urbanpatch dynamics

Application of integration Comparisonsbetween Baltimore and PhoenixHaving laid out a coarse conceptual scheme for inte-grating social and ecological principles and a more spe-cific modeling framework that deals with heterogeneityand dynamics of urban ecological systems we turnnext to a consideration of the distinctly different citieschosen for inclusion in the LTER network In almost everycharacteristic Phoenix and Baltimore are dramatically dif-ferent (Table 1) In terms of the physical environmentsBaltimore is an eastern seaboard metropolis straddling theCoastal Plain and Piedmont provinces whereas Phoenix isan inland city situated in a broad alluvial basin at the con-fluence of two large desert rivers Phoenix has a hot dryclimate (receiving less than 200 mm of precipitationannually) whereas Baltimore has a mesic temperate zoneclimate (1090 mm of precipitation) Natural hydrologicalregimes are flashy (rapid rises and falls in flows) in

Phoenix whereas in Baltimore rainfall and runoff aremore uniform throughout the year The transport of waterand materials by surface runoff is thus highly episodic inPhoenix where baseflow exists only in manmade canalsand as treated sewage effluent Baltimorersquos hydrology andmaterial transport in contrast can be studied using stan-dard watershed approaches although flashiness associatedwith urbanization dictates a focus on storm flows eventhere

Ecological contrasts between Baltimore and Phoenixalso are striking deciduous forest versus desert high(temperate forest) versus low (desert) biotic diversity anddifferent disturbances that initiate succession (whichprobably occurs at very different rates) Water is undoubt-edly the primary limiting factor to production in thedesert region that Phoenix occupies although nitrogenlimitation also is prevalent in both terrestrial and aquaticecosystems (Grimm and Fisher 1986) In Baltimorersquos east-ern deciduous forest limiting factors vary seasonally butmay include low temperatures and frost in winter and soilnutrient limitation and occasional late summer droughtduring the growing season

In terms of societal organization there are importantdifferences as well Modern Phoenix is a much younger citythan Baltimore having been established only after the Civ-il War and having experienced a meteoric rise in popula-tion just since World War II (and interestingly since theinvention of air conditioning) Baltimore is more than 300years old it was established as a seaport after a long histo-ry of scattered habitation of the area (see Foresman et al1997 for a detailed analysis of land-use change in theregion) The site of Baltimore did not support a large pre-historic urban settlement but the Hohokam civilization in

July 2000 Vol 50 No 7 BioScience 581

Articles

Table 1 Contrasting characteristics of Baltimore and Phoenix two metropolises that are recent additions to the US Long-Term Ecological Research (LTER) network

Characteristic Baltimore Phoenix

Climate Mesic temperate Arid hotGeographicndashgeomorphic Landndashsea margin Alluvial basinTopography Coastal plain Flat outcrops and buttes surrounding mountainsHydrologic characteristics Seasonal runoff systems Flashy episodic surface runoffNatural vegetation Eastern deciduous forest Sonoran desert scrubNative biotic diversity High LowndashmoderateInvasibilityndashimpact of exotic biota Moderate HighPrimary limiting factor (ecological) Seasonally variable WaterNatural disturbances Hurricane Fire flash floodSuccession rate Moderate SlowRemnant patches Forest DesertPrehistory Low density scattered Large civilization 1000 years before presentHistory Early seaport Abandoned until 100 years before presentAge of city 300+ years Less than 100 yearsRate of population growth Moderate RapidUrban growth mode Spread and redevelopment SpreadUrban form Compact with core and fringing suburbs Extreme spread coalescing cities interior open

spaceLimits to expansion None Public and Indian landInterior open space Abandoned Never developed or remnantEconomic base Industrial Hi-tech tourismEcosystem boundaries of LTER study Primary Statistical Metropolitan Area County or regulated portion of watershedPatch definition Watershed socioeconomic patches Combination of patch age position neighbors land

ecological patches use land-use history

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

central Arizona included thousands of inhabitants until itsdemise (circa 1350 AD)

Because of climatic conditions the Phoenix area wasnot actively settled until quite late in prehistory (circa 500AD) compared with surrounding regions of the South-west The limitation was insufficient rainfall for agricul-ture hence any kind of substantial occupation wouldrequire knowledge of irrigation techniques and socialorganization to build and maintain the system TheHohokam civilization was able to work cooperatively andestablish a settlement system centered in the valley whichgrew quickly From prehistoric times to the present theenvironmental challenges presented by an arid environ-ment have required cooperative activity by human groupsfor urban centers in such environments to succeed Indi-vidual farmsteads such as those that characterized Balti-morersquos prehistory would have been at serious risk in anarid environment In contrast prehistoric settlement ofthe Baltimore area succeeded based on small-scale agricul-ture and harvesting of coastal resources by small socialgroups However for Baltimore to move beyond its small-scale roots as it has done in historic times cooperativeaction was required this time to build and maintain theharbor and to establish global trading connections Thesummons to cooperative action was both environmental(an accessible harbor) and socioeconomic (a market forlong-distance exchange of goods) The establishment of aplantation culture both relied on and contributed to glob-al trade because of the perceived need for slave labor

Finally present-day contrasts in population growth andcharacteristics of urban growth and form reflect the dis-tinction between an older city that has undergone eco-nomic restructuring (loss of manufacturing jobs and moreservice jobs) and a newer city with an economic base thathas always been primarily in the service sector withrecently high-tech manufacturing The Phoenix metro-politan area is one of the fastest growing in the nation(exceeded only by Las Vegas) with a population that isprojected to double within approximately 25 years (Hall etal 1998) and spreading development that consumes up to4 square miles of desert or agricultural land per 1000 res-idents added to the population (Gober et al 1998) Thepopulation of Baltimore city has declined 23 between1950 and 1990 but growth is high at the county levelreflecting flight from urban to suburban and rural com-munities (Rusk 1996)

Despite these marked dissimilarities we see an oppor-tunity for convergence in the realm of urban ecologicalstudy For both projects the central objective is to under-stand how land-use change over the long term influencesecological pattern and process and how this suite ofchanges feeds back through the social system to drive fur-ther change (eg Figure 3) Furthermore although someof the key questions posed for the two systems are neces-sarily different there are some common questions and acommon approach has been adopted for answering many

of them The guiding principle of the LTER programmdashthat research should whenever possible enable compari-son between different ecosystems and across sitesmdashdic-tates attention to ensuring comparability of approachSome of the research efforts of the Baltimore and Phoenixstudies reveal how social and ecological research can beintegrated in ways that permit cross-comparison of results

Baltimore Ecosystem Study Integration ofphysical biological and social drivers of water-shed dynamics The Baltimore study uses watersheds asfundamental units in which to study the reciprocal inter-actions of the social biophysical and built environmentsOver the past 300 years human settlement and land man-agement have substantially changed the character andproductivity of the Chesapeake Bay the largest and mostproductive estuarine system in the world (Brush 1994)Hydrologists ecologists and social scientists togetherwith public agencies nonprofit organizations and com-munity groups are working to understand how people atdifferent scales (households neighborhoods and munici-palities) directly and indirectly affect water quality in thewatersheds of the Baltimore metropolitan region Initialresearch has shown a significant relationship between con-centration of political and economic power in the city andthe different levels of public and private investment ingreen infrastructure among neighborhoods (green openspaces and trees) Additional research is focusing on thedirect ways in which households might affect water quali-ty through irrigation and through the use of fertilizers andpesticides as well as on how such land-management prac-tices vary with household demographic and socioeconom-ic characteristics

This research will provide important information onhow people influence urban watersheds at different scalesand will result in a hydrologicalndashecologicalndashsocial water-shed model that policymakers planners and managerscan use to assess strategies for improving the water quali-ty of the watersheds

Central ArizonandashPhoenix study Urban fringedynamics Phenomenal rates of urban growth andexpansion in the Phoenix metropolitan area are beingstudied with the intention of defining a new frameworkthrough which to view the ecosystem (Gober et al 1998)The doubling of population in each of the last two 30-yearperiods has led to a rapid spread of the Phoenix urban areainto former farmlands and undisturbed desert landscapesTo monitor this growth researchers are analyzing data onthe exact locations of new residences for each year of thepast decade The data reveal that almost all new single-family residences are built along the periphery of the cityand that urban sprawl acts as a ldquowave of advancerdquo thatspreads out from several nodes of urban development ashas been observed for other youthful cities (Whyte 1968)The speed of this wave and its geographic dimensions

582 BioScience July 2000 Vol 50 No 7

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

seem to respond to conditions of the local economy andcharacteristics of the landscape In turn the landscape andlocal ecosystem are transformed by this advance of newhousing in several predictable stagesmdashland-surface prepa-ration (removal of vegetation soil disturbance) infra-structure construction scattered ldquopioneerrdquo housing devel-opments and fill-in of vacant land with housing Behindthe wave neighborhoods age leading to continued trans-formations in the nature of the human and biotic popula-tions that inhabit those spaces This analysis provides alocational tool that is more sensitive to the key processesthat define the urban phenomenon than the normal gridmap of the city

Summary and synthesisIf Vitousek et al (1997) are correct that by excludinghumans we cannot possibly understand ecosystems and ifit is critical that the social behavioral and economic sci-ences join in the endeavor to understand ecosystems thenit is essential that ecologists welcome the approaches andmodels of the social behavioral and economic sciencesWe have argued that one way to do so is by focusing on fivenew core topic areas for social science We have argued aswell that the study of cities as ecological systems presentsopportunities for theoretical advances in ecology and thatsuch advances cannot be accomplished without integra-tion of the social sciences

We believe that the expansion of social science activitieswithin the LTER network will do much to facilitate con-struction of the concepts most appropriate for under-standing change in the world as will new initiatives instudies of humans as components of ecosystems With thiscommitment ecologists can begin to ask important ques-tions about interdisciplinary research agree aboutmethodology and measurement and successfully integratesocial and ecological data across scales of time and spaceAt no time has there been a more compelling need forintegration and such a wide diversity of researchers readyto begin

AcknowledgmentsThis paper evolved from discussions among the authors ofconcepts and approaches developed in each of the urbanLTER proposals We thank the many investigators fromboth the Central ArizonandashPhoenix and Baltimore Ecosys-tem Study LTER programs who contributed to the devel-opment of those proposals We particularly wish to thankStuart Fisher for many helpful discussions of the ideas pre-sented herein The ldquocommandmentsrdquo of long-term socialscience research arose from our discussions at a Coordi-nating Committee meeting of the LTER network and weresharpened through the contributions of Steve CarpenterTed Gragson Craig Harris Tim Kratz Pete Nowak andChris Vanderpool and by discussions with Diane HopeMark Hostetler Kimberly Knowles-Yanez and NancyMcIntyre Figures 1 and 2 are adapted and expanded from

diagrams created by Tim Kratz we thank him for provid-ing a framework in which to place those ideas Manythanks to Peter McCartney for his assistance in assemblingFigure 5 Comments from Rebecca Chasan Bill FaganJianguo Wu Weixing Zhu and two anonymous reviewersgreatly improved the manuscript The Baltimore Ecosys-tem Study and Central ArizonandashPhoenix LTER programsare supported by the National Science Foundationrsquos (NSF)Long-Term Studies Program (grant numbers DEB9714835 and DEB 9714833 respectively) The BaltimoreEcosystem Study also acknowledges the EnvironmentalProtection AgencyndashNSF joint program in Water andWatersheds project number GAD R825792 and theUSDA Forest Service Northeastern Research Station forsite management and in-kind services Uninterruptedtime for manuscript completion was afforded N B G as avisiting scientist at the National Center for EcologicalAnalysis and Synthesis (a center funded by NSF grant noDEB-94-2153 the University of CaliforniandashSanta Barbarathe California Resources Agency and the California Envi-ronmental Protection Agency)

References citedAmalfi FA 1999 Sudden impact Monsoons and watershed biota cause

rapid changes at Tempe Town Lake Arizona Riparian CouncilNewsletter 12 1 3ndash4

Baker VR 1977 Stream channel response to floods with examples fromcentral Texas Geological Society of America Bulletin 88 1057ndash1071

Borden S 1993 The human component of ecosystems Pages 72ndash77 inMcDonnell MJ Pickett STA eds Humans and Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Bormann FH Likens GE 1967 Nutrient cycling Science 155 424ndash429Botsford LW Castilla JC Peterson C 1997 The management of fisheries

and marine ecosystems Science 277 509ndash515Boyden S Millar S Newcombe K OrsquoNeill B 1981 The Ecology of a City and

its People The Case of Hong Kong Canberra (Australia) AustralianNational University Press

Boyle CA Lavkulich L Schreier H Kiss E 1997 Changes in land cover andsubsequent effects on Lower Fraser Basin ecosystems from 1827 to1990 Environmental Management 21 185ndash196

Brush GS 1994 Human impact on estuarine ecosystems An historicalperspective Pages 397ndash416 in Roberts N ed Global EnvironmentalChange Geographical Perspectives New York Blackwell

Butzer KW 1996 Ecology in the long view Settlement agrosystem strate-gies and ecological performance Journal of Field Archaeology 23141ndash150

Carpenter S Brock W Hanson P 1999 Ecological and social dynamics insimple models of ecosystem management Conservation Ecology 3 (2)4 Available at wwwconsecolorgvol3iss2art4

Chapin FS III Walker BH Hobbs RJ Hooper DU Lawton JH Sala OETilman D 1997 Biotic control over the functioning of ecosystems Sci-ence 277 500ndash504

Clay G 1973 Close Up How to Read the American City New YorkPraeger

Ehrlich P 1997 A World of Wounds Ecologists and the Human DilemmaOldendorfLuhe (Germany) Ecology Institute

Flores A Pickett STA Zipperer WC Pouyat RV Pirani R 1997 Applicationof ecological concepts to regional planning A greenway network forthe New York metropolitan region Landscape and Urban Planning 39295ndash308

July 2000 Vol 50 No 7 BioScience 583

Articles

584 BioScience July 2000 Vol 50 No 7

Articles

Folke CA Jansson J Costanza R 1997 Ecosystem appropriation by citiesAmbio 26 167ndash172

Foresman TW Pickett STA Zipperer WC 1997 Methods for spatial andtemporal land use and land cover assessment for urban ecosystems andapplication in the greater BaltimorendashChesapeake region UrbanEcosystems 1 210ndash216

Gober P Burns EK Knowles-Yanez K James J 1998 Rural to urban landconversion in metropolitan Phoenix Pages 40ndash45 in Hall JS Cayer NJWelch N eds Arizona Policy Choices Tempe (AZ) Arizona State Uni-versity Morrison Institute for Public Policy

Grimm NB 1997 Opportunities and challenges in urban ecologicalresearch Proceedings of the International LTER meeting 10ndash14 Nov1997 Taipei Taiwan

Grimm NB Fisher SG 1986 Nitrogen limitation in a Sonoran Desertstream Journal of the North American Benthological Society 5 2ndash15

Groffman PM Likens GE eds 1994 Integrated Regional Models Interac-tions between Humans and their Environment New York Chapman ampHall

Grossmann WD 1993 Integration of social and ecological factorsDynamic area models of subtle human influences on ecosystems Pages229ndash245 in McDonnell MJ Pickett STA eds Humans and Compo-nents of Ecosystems The Ecology of Subtle Human Effects and Popu-lated Areas New York Springer-Verlag

Grove JM Burch WR Jr 1997 A social ecology approach and applicationsof urban ecosystem and landscape analyses A case study of BaltimoreMaryland Urban Ecosystems 1 259ndash275

Hall JS Cayer NJ Welch N eds 1998 Arizona Policy Choices Tempe (AZ)Arizona State University Morrison Institute for Public Policy

Jacobs J 1962 The Death and Life of Great American Cities The Failure ofTown Planning New York Random House

Likens GE Bormann FH 1995 Biogeochemistry of a Forested Ecosystem2nd ed New York Springer-Verlag

MacLeish WH 1994 The Day before America Changing the Nature of aContinent New York Henry Holt

Matson PA Parton WJ Power AG Swift MJ 1997 Agricultural intensifica-tion and ecosystem properties Science 277 504ndash509

McDonnell MJ Pickett STA 1990 Ecosystem structure and function alongurbanndashrural gradients An unexploited opportunity for ecology Ecol-ogy 71 1231ndash1237

______1993 Humans as Components of Ecosystems The Ecology of Sub-tle Human Effects and Populated Areas New York Springer-Verlag

McDonnell MJ Pickett STA Pouyat RV 1993 The application of the eco-logical gradient paradigm to the study of urban effects Pages 175ndash189in McDonnell MJ Pickett STA eds Humans as Components ofEcosystems The Ecology of Subtle Human Effects and PopulatedAreas New York Springer-Verlag

McPherson EG Nowak D Heisler G Grimmond S Souch C Grant RRowntree R 1997 Quantifying urban forest structure function andvalue The Chicago urban forest climate project Urban Ecosystems 149ndash61

Morgan T 1993 Wilderness at Dawn The Settling of the North AmericanContinent New York Simon amp Schuster

Noble IR Dirzo R 1997 Forests as human-dominated ecosystems Science277 522ndash525

OrsquoHara SL Street-Perrott FA Burt TP 1993 Accelerated soil erosionaround a Mexican lake caused by prehispanic agriculture Nature 36248ndash51

Padoch C 1993 Part II A human ecologistrsquos perspective Pages 303ndash305 inMcDonnell MJ Pickett STA eds Humans as Components of Ecosys-tems The Ecology of Subtle Human Effects and Populated Areas NewYork Springer-Verlag

Pickett STA McDonnell MJ 1993 Humans as components of ecosystemsA synthesis Pages 310ndash316 in McDonnell MJ Pickett STA edsHumans as Components of Ecosystems The Ecology of Subtle HumanEffects and Populated Areas New York Springer-Verlag

Pickett STA Parker VT Fiedler P 1992 The new paradigm in ecologyImplications for conservation biology above the species level Pages65ndash88 in Fiedler P Jain S eds Conservation Biology The Theory and

Practice of Nature Conservation Preservation and Management NewYork Chapman amp Hall

Pickett STA Burch WR Jr Dalton SE Foresman TW Grove JM RowntreeR 1997 A conceptual framework for the study of human ecosystemsin urban areas Urban Ecosystems 1 185ndash199

Pickett STA Wu J Cadenasso ML 1999 Patch dynamics and the ecologyof disturbed ground A framework for synthesis Pages 707ndash722 inWalker LR ed Ecosystems of the World Ecosystems of DisturbedGround Amsterdam Elsevier Science

Redman CL 1992 The impact of food production Short-term strategiesand long-term consequences Pages 35ndash49 in Jacobsen JE Firor J edsHuman Impact on the Environment Ancient Roots Current Chal-lenges Boulder (CO) Westview Press

______ 1999 Human Impact on Ancient Environments Tucson (AZ)University of Arizona Press

Rees WE Wackernagel M 1994 Ecological footprints and appropri-ated carrying capacity Measuring the natural capital require-ments of the human economy Pages 362ndash390 in Jansson AMHammer M Folke C Costanza R eds Investing in Natural Capi-tal The Ecological Economics Approach to Sustainability Wash-ington (DC) Island Press

Reiners WA 1986 Complementary models for ecosystems American Nat-uralist 127 59ndash73

Rice DS 1996 Paleolimnological analysis in the Central Peteacuten GuatemalaPages 193ndash206 in Fedick SL ed The Managed Mosaic Ancient MayaAgricultural and Resource Use Salt Lake City (UT) University of UtahPress

Rusk D 1996 Baltimore Unbound Baltimore Abell Foundation and JohnsHopkins University Press

Russell EWB 1993 Discovery of the subtle Pages 81ndash90 in McDonnell MJPickett STA eds Humans as Components of Ecosystems The Ecologyof Subtle Human Effects and Populated Areas New York Springer-Verlag

Stringer C McKie R 1996 African Exodus The Origins of ModernHumanity New York Henry Holt

Sukopp H 1990 Urban ecology and its application in Europe Pages 1ndash22in Sukopp H Hejny S Kowarik I eds Urban Ecology Plants and PlantCommunities in Urban Environments The Hague (The Netherlands)SPB Academic Publishers

Sukopp H Werner P 1982 Nature in Cities A Report and Review of Stud-ies and Experiments Concerning Ecology Wildlife and Nature Con-servation in Urban and Suburban Areas Strasbourg (France) Councilof Europe

Tansley AG 1935 The use and abuse of vegetational concepts and termsEcology 16 284ndash307

Turner BL II Meyer WB 1993 Environmental change The human factorPages 40ndash50 in McDonnell MJ Pickett STA eds Humans as Compo-nents of Ecosystems The Ecology of Subtle Human Effects and Popu-lated Areas New York Springer-Verlag

Turner BL II Clark WC Kates RW Richards JF Matthews JT Meyer WBeds 1990 The Earth as Transformed by Human Action Global andRegional Changes in the Biosphere over the Past 300 Years New YorkCambridge University Press

[UN] United Nations Population Division 1997a Urban and Rural Areas1950ndash2030 (the 1996 revision) New York United Nations

______ 1997b Urban Agglomerations 1950ndash2015 (the 1996 revision)New York United Nations

van Andel TH Zangger E Demitrack A 1990 Land use and soil erosion inprehistoric and historical Greece Journal of Field Archaeology 17379ndash396

Vitousek PM Mooney HA Lubchenco J Melillo J 1997 Human domina-tion of Earthrsquos ecosystem Science 277 494ndash499

Whyte WH 1968 The Last Landscape Garden City (NY) DoubledayWu J 1999 Hierarchy and scaling Extrapolating information along a scal-

ing ladder Canadian Journal of Remote Sensing 25 367ndash380Wu J Loucks OL 1995 From balance of nature to hierarchical patch

dynamics A paradigm shift in ecology Quarterly Review of Biology70 439ndash466