ISSN 1946-7664. MCFNS 2013 AVAILABLE ONLINE AT HTTP://MCFNS.COM Submitted: Apr. 30, 2013Vol. 5, Issue 2, pp. 151–169 Accepted: Jul. 19, 2013Mathematical and Computational Published: Sep. 30, 2013Forestry&Natural-Resource Sciences Last Correction: Sep. 3, 2013

QUANTITATIVE METRICS FOR ASSESSING PREDICTEDCLIMATE CHANGE PRESSURE ON NORTH AMERICAN TREE

SPECIES

Kevin M Potter1, William W Hargrove2

1Department of Forestry and Environmental Resources, North Carolina State University, Research Triangle Park, NC, USA2USDA Forest Service, Southern Research Station, Asheville, NC, USA

Abstract. Changing climate may pose a threat to forest tree species, forcing three potential population-level responses: toleration/adaptation, movement to suitable environmental conditions, or local extirpation.Assessments that prioritize and classify tree species for management and conservation activities in the faceof climate change will need to incorporate estimates of the risk posed by climate change to each species.To assist in such assessments, we developed a set of four quantitative metrics of potential climate changepressure on forest tree species: (1) percent change in suitable area, (2) range stability over time, (3) rangeshift pressure, and (4) current realized niche occupancy. All four metrics are derived from climate changeenvironmental suitability maps generated using the Multivariate Spatio-Temporal Clustering (MSTC)technique, which combines aspects of traditional geographical information systems and statistical clusteringtechniques. As part of the Forecasts of Climate-Associated Shifts in Tree Species (ForeCASTS) project, wecalculated the predicted climate change pressure statistics for North American tree species using occurrencedata from the USDA Forest Service Forest Inventory and Analysis (FIA) program. Of 172 modeled treespecies, all but two were projected to decline in suitable area in the future under the Hadley B1 GlobalCirculation Model/scenario combination. Eastern species under Hadley B1 were predicted to experience agreater decline in suitable area and less range stability than western species, although predicted range shiftdid not differ between the regions. Eastern species were more likely than western species, on average, tobe habitat generalists. Along with the consideration of important species life-history traits and of threatsother than climate change, the metrics described here should be valuable for efforts to determine whichspecies to target for monitoring efforts and conservation actions.

Keywords: climate change; range shift pressure; risk assessment; multivariate clustering; human-assisted migration; niche occupancy; forest health monitoring; conservation.

1 Introduction

The forests of the United States are expected to expe-rience extensive ecological, social and economic effects asa result of climate change (Malmsheimer et al., 2008).Specifically, forest ecosystem functions and attributesare likely to be altered as a result of climatic changes(Stenseth et al., 2002) that are forecast to include anincrease in mean surface temperatures of 2 ◦C to 4.5◦C, longer growing seasons, and changes in temporaland spatial precipitation patterns (International Panelon Climate Change, 2007). Climate change is likely topose a severe threat to the viability of forest tree speciesthemselves, which will be forced either to adapt to new

conditions or to shift their ranges to more favorableenvironments. Evidence suggests that tree species arecurrently exhibiting changes in distribution and phenol-ogy in response to climate change (Parmesan and Yohe,2003; Root et al., 2003; Woodall et al., 2009; Zhu et al.,2012). As species move poleward or upslope in the faceof climate change, some species will likely disappear orbe restricted to isolated refugia, while others may ex-pand greatly (Iverson and McKenzie, 2013). Biologistshave expressed concerns that species may be extirpatedas their access to suitable habitat decreases (Thomaset al., 2004). The growth and survival of forest treespecies will depend on the maintenance of suitable habi-tats and on the availability of genotypes for coloniza-

Copyright © 2013 Publisher of theMathematical and Computational Forestry & Natural-Resource SciencesPotter and Hargrove (2013) (MCFNS 5(2):151–169). Manuscript Editor: Greg C Liknes

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 152

tion of those habitats (Rehfeldt, 1999). Managers anddecision-makers will need tools to assess the potentialimpacts of climate change on the broad diversity of for-est tree species across North America and elsewhere.

As environmental changes push the habitat of plantspecies out of their climatic tolerance limits, species mayrespond by adapting to the new conditions, by shiftingvia migration to suitable environmental conditions, or bybecoming locally extirpated (Davis et al., 2005). Adap-tation via natural selection may be unlikely for foresttree species in many cases, given their long generationtimes (Rehfeldt et al., 1999; St. Clair and Howe, 2007),although some tree species may be able to evolve quicklyin the face of new environmental conditions (Petit et al.,2004). Much innovative work has predicted the futuredistribution of habitat suitability for forest tree speciesunder climate change (Iverson et al., 2004a, 2004b, 2008;Matthews et al., 2011; Rehfeldt et al., 2006; Schwartz etal., 2006), although the results of such efforts typicallydo not account for whether tree species can span, with-out human assistance, the distances expected betweenlocations with suitable conditions currently and thosewith suitable conditions in the future. Indeed, manytree species successfully migrated long distances follow-ing the most recent cold period of the Pleistocene, butthere is concern that these same species may not be ableto match the much more rapid climate shifts expectedin the near future (Davis and Shaw, 2001).

Areas within the distributions of forest tree species arelikely to experience different degrees of climate changepressure. For example, the paleorecord suggests thatpopulations at the trailing edge of a species’ shiftingdistribution were often extirpated, resulting in a lati-tudinal displacement of range rather than a simple ex-pansion into newly favorable region (Davis and Shaw,2001). Already, a disproportionate number of popula-tion extinctions have been documented along southernand low-elevation range edges in response to recent cli-mate warming, resulting in contraction of species’ rangesat these warm boundaries (Parmesan, 2006). At thesame time, these trailing edge populations appear tohave played a key role for the maintenance of biodi-versity through the Quaternary, and Hampe and Petit(2005) argue that rear-edge populations are dispropor-tionately important for the long-term conservation ofgenetic diversity, phylogenetic history and evolutionarypotential of species.

Assessments that prioritize and classify tree speciesfor management and conservation activities in the faceof climate change (e.g., Aubry et al., 2011; Devine etal., 2012; Matthews et al., 2011) need to incorporate es-timates of the risk climate change poses to each species.To assist in such assessments, we developed a set offour quantitative metrics of predicted climate change

pressure on forest tree species: (1) percent change insuitable area, (2) range stability over time, (3) rangeshift pressure, and (4) current realized niche occupancy.All four metrics are derived from climate change envi-ronmental suitability maps generated using the Multi-variate Spatio-Temporal Clustering (MSTC) technique(Hargrove and Hoffman, 2005). Combining aspects oftraditional geographical information systems (GIS) andstatistical clustering techniques, MSTC can be used tostatistically predict environmental niche envelopes toforecast a species’ potential geographic range under al-tered environmental conditions such as those expectedunder climate change (Hargrove and Hoffman 2003).Global in scope, it incorporates 16 spatial climate, soils,and geomorphology variables, and generates maps ata resolution of 4 km2. The advantages of the MSTCtechnique include its capacity to easily generate climatechange environmental suitability maps for a large num-ber of species, its relatively high-resolution results appli-cable at the population level, and its ability to predictsuitable habitat globally (for species potentially movingfrom Mexico to the United States, for example, or fromthe United States to Canada).

As a part of the Forecasts of Climate-Associated Shiftsin Tree Species (ForeCASTS) project (Potter et al.,2010), we calculated the climate change pressure met-rics for 172 North American tree species using occur-rence data from the USDA Forest Service Forest Inven-tory and Analysis (FIA) program as training occurrencelocations. FIA data are an unparalleled source of treelocation information in the United States because theFIA program maintains a national network of approxi-mately 125,000 fixed-area forested plots from which treeinventory data are collected in a consistent manner andon a regular basis (Woudenberg et al., 2010). Becauseof the large number of plots and because each plot rep-resents a little more than 2,400 ha of forest (Bechtoldand Patterson, 2005), the FIA data reliably representthe general extent of common tree species.

2 Materials and Methods

2.1 Tree occurrence data We generated metrics ofprojected climate change pressure for 172 North Amer-ican forest tree species (Appendix A, Tab. 4). Becausewe used inventory data collected by the USDA For-est Service as training occurrence locations for our treespecies climate projections, and because we wanted toprovide range-wide estimates of climate change pressureon species, we included only species for which more than75 percent of the estimated range area occurs within theborders of the conterminous 48 United States, based onE.L. Little’s range maps (Little, 1971; United States Ge-ological Survey, 1999).

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 153

To select climate change training data for eachof our study species, we used coarsely georeferencedspecies occurrence data available from the USDAForest Service Forest Inventory and Analysis (FIA)program (Woudenberg et al., 2010), available athttp://www.fia.fs.fed.us/tools-data/. The FIA programis the primary source of information about the extent,condition, status, and trends of forest resources acrossall ownerships in the United States (Smith, 2002). FIAapplies a nationally consistent sampling protocol using aquasi-systematic design to conduct a multi-phase inven-tory of all forested land ownerships; the national sam-ple intensity is approximately one plot per 2,428 ha ofland (Bechtold and Patterson, 2005). It maintains asystem of approximately 125,000 fixed-area (each ap-proximately 0.067 hectares) inventory plots on accessibleforested land across the 48 conterminous United Statesand southeastern Alaska; field crews collect data on morethan 300 variables, including forest type, tree species,tree size and tree condition (Smith, 2002; Woudenberget al., 2010). The plots consist of four, 7.2-m fixed-radiussubplots spaced 36.6 m apart in a triangular arrange-ment with one subplot in the center (Woudenberg et al.,2010). All trees with a diameter at breast height (dbh)of at least 12.7 cm are inventoried on forested subplots.The FIA system is designed so that field crews revisitplots in the eastern United States every five years, with20 percent of all plots remeasured every year on a 5-yearrotating basis. In the western United States, 10 percentof plots are remeasured every year on a 10-year rotat-ing basis. Initial annual inventory plots were establishedbetween 1999 and 2005. All inventory data are publiclyavailable. By law, the exact coordinates of FIA plots areslightly altered to protect the privacy of forest landown-ers, with most of the adjusted coordinates located within0.8 km (0.5 miles) and all within 1.61 km (1 mile) ofthe actual plot coordinates. Additionally, some privateplot coordinates may be “swapped” with those of an-other private plot within the same county with similarattributes, such as forest type, stand-size class, latitudeand longitude (Woudenberg et al., 2010). Obscuring theoriginal plot coordinates should have little effect on theresults of this study given the resolution of the analysisand measures undertaken to avoid overtraining the data.

In some cases, we combined multiple FIA speciescodes into a single species group when doing so was taxo-nomically justified (e.g., combining all hawthorn species,which are difficult to differentiate even by experts, into asingle “Crataegus spp.” category). We excluded specieswhich occurred on fewer than 10 FIA plots, to ensureadequate sampling. We wanted to include only plotswhere a given tree species has been able to attain repro-ductive maturity, so an FIA plot was used as a train-ing occurrence location for a large tree species if it con-

tained at least one tree greater than 25.4 cm dbh or9.14 m in height (class 1), and for a smaller tree speciesif it contained a tree at least 12.7 cm dbh or 6.1 m inheight (class 2). For the smallest tree species, includ-ing those that often occur in a shrubby form, such asPrunus americana and Quercus gambelii, plots were in-cluded if they contained an individual at least 6.35 cmdbh or 3.048 m in height (class 3). The same was true forAmerican chestnut (Castanea dentata), a species thathas been decimated by an exotic fungal disease causedby Cryphonectria parisitica (Loo, 2009; Russell, 1987),and which continues to exist in upland hardwood forestsof the eastern United States in the form of sprouts fromblight-killed trees (Stephenson et al., 1991). Each of the172 species was classified as western or eastern, witheastern species subdivided into northern, southern andgeneral species, as in Woodall et al. (2009).

2.2 Predictions of future habitat suitabilityMSTC is a technique that employs non-hierarchical clus-tering to classify GIS raster cells with similar environ-mental conditions into categories (Hargrove and Hoff-man, 2005). It uses the normalized values of each envi-ronmental condition for every raster cell as a set of co-ordinates that together specify a position for that cell ina data space having a separate dimension for each of theenvironmental characteristics. Normalization gives envi-ronmental parameters measured in different units equalspacing by establishing a mean of zero and unit standarddeviation (Hargrove and Hoffman, 2005). Two cells fromanywhere on the map with similar combinations of envi-ronmental characteristics will be located near each otherin this data space. Their proximity and relative posi-tions in data space will quantitatively reflect their en-vironmental similarities, allowing these cells to be clas-sified into groups or “ecoregions” with other cells pos-sessing similar environmental conditions; each ecoregioncontains roughly an equal amount of multivariate envi-ronmental heterogeneity (Hargrove and Hoffman, 2005;Hoffman et al., 2005). The MSTC process generates out-put maps which group and display each pixel as part ofan “ecoregion” with other pixels possessing similar en-vironmental conditions. It is possible to choose whetherthe map contains many small ecoregions, each containinglittle environmental heterogeneity, or only a few ecore-gions, each containing a relatively large amount of en-vironmental heterogeneity. The results presented herewere generated using a fine division of 4 km2 (1.25 ar-cmin) pixels globally into 30,000 ecoregions, each witha relatively small amount of environmental heterogene-ity. This is the finest resolution at which global data areavailable; the MSTC method has been applied at a finerresolution when appropriate input data were available(e.g., Hargrove and Hoffman, 2003, 2005).

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 154

Global in scope, this MSTC analysis incorporates 16spatial bioclimatic, topographic, and edaphic environ-mental variables (Tab. 1) and generates maps using geo-referenced occurrence information as training data for agiven species (Potter and Hargrove, 2012). These vari-ables were used because they play an important role indetermining the geographic distribution of plants acrosslarge areas (Lugo et al., 1999; Neilson, 1995; Prentice etal., 1992). Climatic data were custom downscaled fromHijmans et al. (2005), topographic variables were cus-tom downscaled from Moore et al. (1991), and edaphicdata were custom downscaled from the Global Soil DataTask Group (2000). Saxon et al. (2005) and Baker etal. (2010) provide additional details about the variablesused in the MSTC analysis, including how they weredownscaled. Using the MSTC approach, it is possible toassign a different weight to each of the 16 environmentalvariables, but we gave them all equal weight to allowfor a rapid assessment of climate change pressure acrossthe 172 species and to maintain consistency across thespecies.

Because MSTC can track the same clustered combi-nation of environmental conditions at any location ordate using future climatic forecasts, it has been ap-plied to identify potential climatic refugia predicted byglobal shifts in environmental conditions (Baker et al.,2010; Saxon et al., 2005), to determine quantitativezones for seed transfer that take climate change intoaccount (Potter and Hargrove, 2012), and to statisti-cally model environmental niche envelopes to forecastsuitable habitat conditions for species under altered en-vironmental conditions such as expected under globalclimate change (Hargrove and Hoffman, 2005; Potter etal., 2010). In such situations, future climate projectionsare used, while edaphic and topographic data are heldconstant over time. MSTC is an appropriate tool forthe assessment of the potential genetic effects of climatechange on forest tree species because it is able to rapidlypredict changes in suitable habitat for a large number ofspecies, it allows for flexible occurrence data inputs, itgenerates relatively high-resolution results applicable atthe population level, it has the ability to predict suitablehabitat beyond the borders of the United States, andit incorporates pertinent environmental variables asso-ciated with plant distributions (Potter et al., 2010).

Using FIA plot training occurrence data (Fig. 1),we have produced maps that predict the locationand suitability of current and future environmentalconditions for North American tree species usingthe MSTC technique (Fig. 2). We generated thesemaps and associated climate change pressure metricsfor the ForeCASTS project (Potter et al., 2010),which aims to assess how changing climate con-ditions could affect the genetic integrity of forest

Table 1: The Multivariate Spatio-Temporal Clusteringtechnique employed 16 spatial environmental variables,collected at a resolution of 4 km2, to define the 30,000quantitative global ecoregions used to predict the loca-tion and quality of current and future suitable environ-mental conditions for North American tree species.

Category Spatial environmental variableClimaticvariables

Annual biotemperature (sum ofmonthly mean temperature wheremean ≥ 5 ◦C)Growing season (number of consecu-tive months with mean ≥ 5 ◦C)Mean diurnal temperature range (◦C)Mean precipitation in the coldestquarter (mm)Mean precipitation in the driest quar-ter (mm)Mean precipitation in the warmestquarter (mm)Mean precipitation in the wettestquarter (mm)Mean temperature in the coldest quar-ter (◦C)Mean temperature in the warmestquarter (◦C)Precipitation/potential evapotranspi-ration

Topographicvariables

Annual potential solar insolation(kW/m2)Compound topographic index (rela-tive wetness)

Edaphic Profile available water capacity (mm)variables Soil bulk density (g/cm3)

Total soil nitrogen (g/m2)Total soil carbon (g/m2)

tree species and populations. All the maps andstatistics described here are available for viewing athttp://www.geobabble.org/h̃nw/global/treeranges3/cli-mate change/. At this Web site, a page exists for eachspecies, containing (1) maps of training occurrencelocations, (2) maps of locations with currently suitableenvironmental conditions, (3) maps of locations ex-pected to be suitable under multiple global circulationmodels (GCMs) and emissions scenarios at two timepoints, and (4) maps of minimum required movementunder one GCM/scenario combination for 2050. Whenthey exist, links to corresponding climate change projec-tions from other researchers using different techniques

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 155

Legend

FIA plots with Quercus rubra

Quercus rubra distribution

Legend

FIA plots with Pinus lambertiana

Pinus lambertiana distribution

(a)

(b)

Figure 1: Forest Inventory and Analysis (FIA) plotsused as Multivariate Spatio-Temporal Clustering occur-rence data for (a) northern red oak (Quercus rubra) and(b) sugar pine (Pinus lambertiana). The plot locationsare approximate. Species distributions are digitized ver-sions of E.L. Little’s range maps (Little, 1971; UnitedStates Geological Survey, 1999).

(Crookston, 2013; Prasad et al., 2013) are includedalong with the ForeCASTS climate suitability maps.

To avoid overtraining the results because of relativedifferences in local species abundance, we selected ecore-gions in the current-time suitable environmental con-ditions map based on the geographic distribution andcommonness of the species. For very common species,an ecoregion was included as suitable when it inter-sected with four FIA plots containing that species. Forless common species, the threshold was three occur-rence plots per ecoregion, while it was two occurrenceplots for uncommon species and one occurrence plot

Figure 2: Multivariate Spatio-Temporal Clustering re-sults for northern red oak (Quercus rubra), (a) Q. rubrapredicted environmental suitability comparison for cur-rent conditions and for 2050 under the Hadley global cir-culation model, B1 emissions scenario, and (b) Q. rubradistance to nearest future suitable conditions under theHadley B1 model-scenario combination.

for the rarest species. Species commonness was deter-mined by the number of ecoregions that intersect withtwo FIA occurrence plots containing that species; verycommon species encompassed 343 or more unique ecore-gions, common species encompassed 49 or more and lessthan 343 unique ecoregions, uncommon species encom-passed seven or more and less than 49 ecoregions, andrare species encompassed fewer than seven ecoregions.These training occurrence thresholds were determinedby a simple first-order exponential function.

The maps of currently suitable conditions and of fu-ture predicted environmental conditions also depict ar-eas (ecoregions) of decreasing environmental similarityto the environmental conditions currently present at thetree species training occurrence locations. Defined asthe Euclidean distance in data space between the cen-

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 156

troids of two ecoregion clusters, this ecoregion similarityis displayed on the maps as a grayscale ramp. Darkershades are given to cells belonging to ecoregions moresimilar to any of those that intersect with the trainingoccurrence plots in current time, while lighter shades aregiven to those belonging to less similar ecoregions.

The MSTC approach assumes that trees are optimallyadapted to the environmental conditions existing at theirtraining data locations. This generally holds true forforest tree species (Johnson et al., 2004), although ex-ceptions also exist (Mangold and Libby, 1978; Wu andYing, 2004). It is also important to note that, withsome GCM/emissions scenario combinations, the set ofenvironmental conditions equivalent to those present inan ecoregion in current time may not exist (Fitzpatrickand Hargrove, 2009). If this happens, MSTC will notbe able to predict equivalent locations for that currentecoregion on the future projection maps (Hargrove andHoffman, 2003, 2005). However, these maps may depictfuture locations with environmental conditions that aresimilar to the original current-time ecoregion.

2.3 Metrics of projected climate change pres-sure We used the MSTC mapped results to calculate,for each of the 172 tree species, four metrics of projectedclimate change pressure (percent change in range areaover time, percent range stability over time, range shiftpressure, and current realized niche occupancy). Weused PROC GLM in SAS 9.2 (SAS Institute Inc., 2008)to assess whether the means of the metrics were signifi-cantly different by region and subregion, using a Tukey-Kramer test because of group size differences. We con-ducted a Wilcoxon signed rank test in PROC UNIVARI-ATE to assess to determine whether percent change insuitable habitat area over time was significantly differ-ent than 0 within each group of species (all species, re-gions, and subregions). We also used PROC CORR inSAS 9.2 to test for correlations between each pair ofmetrics, across the entire set of species and within re-gions. We calculated nonparametric Spearman correla-tions, because the variables were not typically normallydistributed, and outliers were present in several cases.Pearson correlations are not likely to be robust in thepresence of non-normality (Kowalski, 1972).

2.3.1 Percent change in suitable area This mea-sure of change over time in environmentally suitable areawas determined by comparing, for each species, the per-cent change in area with suitable environmental con-ditions over time from current conditions to those pro-jected in 2050 under the Hadley B1 GCM/emissions sce-nario combination.

2.3.2 Range stability over time This measure ofrange stability over time was determined by calculat-ing, for each species, the percent of currently suitablearea that remains suitable over time under conditionsprojected in 2050 in the Hadley B1 GCM/emissions sce-nario combination.

2.3.3 Range shift pressure The projected meanshift distance in suitable habitat, or “Minimum RequiredMovement” (MRM) distance, was determined by cal-culating the mean non-zero straight-line distance (mea-sured in grid cells) from each currently suitable 4 km2

raster cell expected to become unsuitable to the nearestexpected suitable habitat cell in 2050 under the HadleyB1 GCM/emissions scenario combination.

2.3.4 Current realized niche occupancy Thebreadth of niches occupied by a species is a strong pre-dictor of extinction risk, with species having narrowniches in general being at greater risk (Brook et al., 2008;Stork et al., 2009). Since the MSTC-derived ecoregionsare quantitatively defined and have fixed environmentalvariability, and since FIA data are sampled in a con-sistent and systematic fashion (Bechtold and Patterson,2005), we used the two in combination to generate a met-ric of current realized niche occupancy. It is calculatedas the number of unique MSTC ecoregions that inter-sect with two or more FIA occurrence plots containinga given species.

3 Results

3.1 Percent change in suitable area Environmen-tally suitable area was projected to decline by an averageof 44.33% across all 172 species in the study by 2050 un-der Hadley B1 (Tab. 2). The decline in suitable area wasprojected to be twice as great in the East as in the West(50.55% compared to 23.82%); this difference was sta-tistically significant. Northern species in the East wereexpected to experience a greater decline in suitable habi-tat area than both southern species and more generallydistributed eastern species in the region, but the differ-ences among the means were not significant. Nationally,September elm (Ulmus serotina) was projected to havethe greatest decline in suitable area, followed by sweetcrabapple (Malus coronaria), chalk maple (Acer leuco-derme), Delta post oak (Quercus similis), and Texasash (Fraxinus texensis) (Appendix A, Tab. 4). All ofthese, with the exception of sweet crabapple, are south-ern species. Only two species were projected to havean increase in suitable habitat: Great Basin bristleconepine (Pinus longaeva) and dwarf live oak (Quercus min-ima). All the 170 other tree species lost projected suit-able habitat. Note, however, that these results are from

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 157

only one GCM/emissions scenario combination and fromone time step, because our emphasis is to describe andillustrate metrics of climate change pressure rather thanto present a range of potential climate change effects.

3.2 Percent range stability over time As withpercent change in suitable habitat over time, the percentof currently suitable habitat expected to remain suit-able was higher, on average, for western species (47.58%) than eastern species (33.66%) (Tab. 2), a differencethat was statistically significant. Range stability wasprojected to be greater for southern and generally dis-tributed species in the East than for northern species,but these differences were not significant. Nationally,mean species range stability was 36.90%. Range stabil-ity values ranged from 0 percent in September elm andless than 2% in chalk maple, Delta post oak, and Ken-tucky coffeetree (Gymnocladus dioicus), to 67.93 percentin loblolly bay (Gordonia lasianthus), 67.9% in bigleafmaple (Acer macrophyllum), 69.28% in Sitka spruce(Picea sitchensis), and 84.82% in dwarf live oak (Ap-pendix A, Tab. 4).

3.3 Range shift pressure Mean minimum requiredmovement (MRM) distance, a measure of range shiftpressure, was projected to be somewhat greater forspecies in the East than in the West (12.62 map cellscompared to 8.97 map cells), although the difference inthese means was not significant. The variation acrossspecies was slightly greater in the West (Tab. 2). Inthe East, northern species were expected to have thegreatest shift pressure, and generally distributed speciesto have the least, with southern species intermediate be-tween the two. Again, however, none of these differenceswas significant. Across all species, the mean shift pres-sure was 11.77 map cells. Expected range shift pressurewas the greatest for September elm (94.78 map cells),Great Basin bristlecone pine (73.34 map cells) and Ken-tucky coffeetree (61.28 map cells), and least for wingedelm (Ulmus alata), post oak (Quercus stellata), cher-rybark oak (Quercus pagoda), and western larch (Larixoccidentalis), all with a mean range shift pressure of ap-proximately 3 map cells or fewer (Appendix A, Tab. 4).

3.4 Current realized niche occupancy Mean re-alized niche occupancy in current time was greater foreastern species than for western species (149.38 uniqueecoregions compared to 85.10, with the means being sig-nificantly different) (Tab. 2). This suggests that easternspecies are more likely than western species, on average,to be habitat generalists. Not surprisingly, generally dis-tributed eastern species intersected with more uniqueecoregions than either northern or southern species,with means significantly different among the subregions.

Southern species had a smaller mean realized niche oc-cupancy than northern species, but the difference wasnot significant. The mean across all species nationallywas 134 unique ecoregions. The range across species wasas low as 1 for Great Basin bristlecone pine and 2 eachin Delta post oak and September elm, and as high as502 for black cherry (Prunus serotina), 518 for greenash (Prunus pennsylvanica), 522 for American elm (Ul-mus americana), and 635 for red maple (Acer rubrum)(Appendix A, Tab. 4).

3.5 Relationship among metrics Nationally, allpairs of metrics were significantly correlated (Tab. 3).The strongest correlation was between percent changein suitable area and range stability (r = 0.833). Sincenearly all species were projected to experience a decreasein suitable area, this shows that species with the smallestdecrease in suitable area had the greatest range stabil-ity. The relationships both between shift pressure andpercent change in suitable area (r = -0.487) and be-tween shift pressure and range stability were negative (r= -0.655); in other words, species were likely to experi-ence less climate change shift pressure (that is, distanceto the nearest suitable projected habitat for 2050 underthe Hadley B1 GCM-scenario combination) with less ofa projected loss of habitat and with a greater amount ofhabitat remaining constant over time. Current realizedniche occupancy, a measure of whether a species is ahabitat generalist or a habitat specialist, was positivelycorrelated with range stability (r = 0.458), but nega-tively correlated with shift pressure (r = -0.505). Thissuggests that habitat generalists, which have been foundto exist in a wider variety of environmental niches andacross large geographical ranges (Fridley et al., 2007),should be able to continue to exist across a broader areawhile not having to move as far to reach future suitablehabitat.

These general patterns existed among eastern speciesas well (Tab. 3), but with stronger correlations. Thepattern was slightly different for western species, how-ever. In the West, a smaller but still significant cor-relation existed between percent change in suitable areaand range stability (r = 0.325). Also, the association be-tween change in suitable area and climate change shiftpressure was not significant, as it was in the easternUnited States and nationally. This suggests that westernspecies projected to lose less overall habitat may need tomove greater distances from currently-suitable/future-unsuitable locations to the nearest future-suitable loca-tions, when they do lose environmentally suitable area.This might be the result of being located on “sky is-lands” (McLaughlin, 1995; Warshall, 1995), which maycause populations occurring mostly at high elevations inthe southern portions of their species ranges (while being

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 158

Table 2: Mean across species values of the climate change pressure metrics developed using the Multivariate Spatio-Temporal Clustering approach, nationally and by region. Percent change is significantly different than 0 for allgroups. Note: EG, eastern-general; EN, eastern-north; ES, eastern-south.

Suitable Area(km2)

RangeStability

ShiftPressure

NicheOccupancy

Region Species Current Future % Change Area (km2) % Mean SD

All 172 971,470 536,995 -44.33 403,500 36.90 11.77 20.80 134.43

East 132 1,144,327 613,065 -50.55 460,407 33.66 12.62 19.41 149.38EG 54 1,761,160 899,699 -53.06 710,863 35.24 10.26 17.91 228.30EN 25 878,119 441,224 -56.59 289,877 25.62 16.54 25.89 123.88ES 53 641,203 402,071 -45.14 285,663 35.85 13.16 17.89 81.00

West 40 401,336 285,962 -23.82 215,706 47.58 8.97 25.36 85.10

Table 3: Spearman correlations between climate changepressure metrics among species, nationally and by re-gion. Correlations are significant at p < 0.05 exceptthose underlined.

% Change Stability Shiftpres-sure

Nicheoccu-pancy

All species% Change . 0.833 -0.487 0.164Stability 0.833 . -0.655 0.458Shift pres. -0.487 -0.655 . -0.505Niche occup. 0.164 0.458 -0.505 .

Eastern species% Change . 0.903 -0. 522 0. 416Stability 0.903 . -0. 657 0.592Shift pres. -0. 522 -0.657 . -0. 643Niche occup. 0.416 0.592 -0.643 .

Western species% Change . 0.325 -0.058 -0.063Stability 0.325 . -0.462 0.643Shift pres. -0.058 -0.462 . -0.427Niche occup. -0.063 0.643 -0.427 .

more broadly dispersed at more northerly latitudes) tohave to move farther to reach an environmentally suit-able location in the future.

4 Discussion

Changing climatic conditions are expected to pose athreat to the viability of forest tree species, which maybe forced either to adapt to new conditions or to shift

their ranges to more favorable environments (Aitken etal., 2008; Davis et al., 2005). Given the limitations infunding available for species-specific management andconservation activities, it will be necessary to comparethe expected impacts of these environmental changesacross multiple species in a region (e.g., Barazani etal., 2008; Coates and Atkins, 2001; Gauthier et al.,2010). We describe quantitative metrics developed forsuch comparative assessments. These are based on cli-mate change map products that combine estimates ofspecies’ edaphic and bioclimatic envelopes with down-scaled products of climate modeling, which can pre-dict the future spatial extent of those environmentalenvelopes and determine where species’ ranges mightexist, disappear, move, grow or shrink under changedclimatic conditions (Harris et al., 2006). Using thesemaps, the metrics describe the existing environmentalvariation across the range of a species (realized currentniche occupancy), the degree to which the area of suit-able environmental conditions is predicted to decreaseor increase over time (percent change in suitable area),the amount of currently suitable area that is expected toremain suitable (range stability over time), and the dis-tance that tree populations currently in areas expectedto become unsuitable would have to travel to reach thenearest suitable location in the future (range shift pres-sure).

Because the main objective of this analysis was to de-scribe a set of predicted climate change pressure met-rics, the results we present here are limited to a singleglobal circulation model/emissions scenario combination(Hadley B1) for a single point in time (2050). Thoroughassessments of risk across forest tree species should likelyconsider multiple GCM/scenario combinations to betteraccount for a range of possible climate change predic-tions. When using downscaled climate predictions from

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 159

the Hadley B1 GCM/scenario combination, we foundthat nearly all of the 172 species in our analysis wereexpected to experience a loss of overall environmentalsuitable area and to need to move a relatively short dis-tance, on average, from newly unsuitable to the near-est future suitable locations (Fig. 3). Generating thesemetrics using other GCMs and scenarios might revealdifferent results.

Figure 3: The 172 species included in the analysis, plot-ted by percent change in environmental suitable areaand range shift pressure (number of grid cells), based onthe Hadley B1 global circulation model/scenario combi-nation. The labeled outlier species are discussed in thetext. Note: EG, eastern-general; EN, eastern-north; ES,eastern-south; W-western.

In the current analysis, the species that are the ex-ceptions from the pattern of suitable area loss with rela-tively small climate change shift pressure are perhapsthe most interesting from a monitoring, managementand conservation perspective. For example, Great Basinbristlecone pine (Pinus longaeva), an extremely long-lived species which occurs in isolated mountain rangesin California, Nevada and Utah (Hiebert and Hamrick,1984), is expected under Hadley B1 to experience an in-crease in suitable area that exceeds 100 percent, whileretaining about 25 percent of its existing suitable habi-tat. The mean distance to suitable future environmentallocations, from areas expected to become unsuitable, isgreater than nearly all other species, however, at about292 km (approximately 73 4-km2 map cells). This isbecause newly suitable locations for the species are pro-jected to develop in Canada, far from its current loca-tions, where much of its current habitat is expected tobecome unsuitable. At the same time, September elm(Ulmus serotina), a species scattered infrequently acrossa few southeastern states (Flora of North America Edito-

rial Committee, 1993+), is projected to lose nearly all itscurrently acceptable habitat, to maintain none of its cur-rently suitable habitat, and to need to shift an averageof 379 km (approximately 95 4-km2 map cells) to reachsuitable conditions in the future. Both species clearlyneed to be closely monitored, and may need to be consid-ered as candidates for facilitated migration (Pedlar et al.,2012; Vitt et al., 2010). Delta post oak (Quercus sim-ilis), Kentucky coffeetree (Gymnocladus dioicus), andCarolina hemlock (Tsuga caroliniana) are other speciesthat may experience particularly intense climate changeimpacts. Meanwhile, dwarf live oak (Quercus minima),from central and northern Florida, is exceptional be-cause, under the Hadley B1 GCM/scenario, it is pro-jected to gain suitable area, maintain about 84 percentof its current suitable area, and need to move only ashort distance from newly unsuitable to newly suitablelocations.

Species in the East are projected, on average, to expe-rience a greater loss of suitable area and a decreased levelof range stability compared to species in the West underHadley B1. Eastern species tend to occur across moreFIA plots than western species and to have larger ar-eas of current environmental suitability. Eastern specieswere inventoried on 2,760 FIA plots on average com-pared to 921 for western species, and had 1,144,237 km2

of current suitable area compared to 401,336 km2. Thesemeans were significantly different. Additionally, easternspecies appear more likely than western species to behabitat generalists than specialists, given the greatermean realized niche occupancy in the East. Thus,greater broad-scale environmental changes may result ingreater loss of suitable area for eastern species, via lati-tudinal displacement of species ranges (Davis and Shaw,2001). Western species, meanwhile, are more likely tobe habitat specialists and to be more limited in theircurrent distributions, perhaps in part because of thegreater topographic complexity of the West. This to-pographic complexity may cause current suitable envi-ronmental conditions for tree species generally to shiftupward in elevation over short enough distances thatgrid cells remain suitable over time, rather than shiftinggreater distances in latitude as in the East. Interestingly,expected shift pressure on species is not significantly dif-ferent between species in the two regions, despite dif-ferences in range stability and change in suitable area.Perhaps most places expected to become unsuitable forwestern species are those at the highest elevations, asa result of the existence of “sky islands” (McLaughlin,1995; Warshall, 1995), from which tree species wouldneed to traverse considerable distances to reach suitablefuture habitat. Though the area encompassed by suchlocations is smaller than that of places in the East ex-pected to become unsuitable, it appears that species in

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 160

both regions would need to span similar distances toreach future suitable environmental conditions.

It is important to note that the projected climatechange pressure metrics we describe here represent onlyone set of inputs necessary for assessments comparingclimate change risk across multiple tree species. Theimpact of climate change on species will vary acrossspecies based on such individualistic traits as seed dis-persal mechanism and life-history strategies (Parmesan,2006; Parmesan and Yohe, 2003), which should be in-corporated into assessments of species susceptibility toclimate change and other threats (Aitken et al., 2008;Myking, 2002; Sjostrom and Gross, 2006). These cli-mate change pressure statistics, along with considera-tion of species’ biological attributes, can allow for assess-ment of whether migrating species, for example, mightbe able to track appropriate environmental conditionsover time and avoid the loss of extensive genetic vari-ation. A loss of important adaptive genetic variationmay be a concern particularly for species that have nar-row habitat requirements, are located exclusively at highelevations, and/or are not able to disperse their propag-ules effectively across long distances. Even if not lo-cally extirpated outright, populations of these and otherspecies could experience significant inbreeding, geneticdrift, and decreased genetic variation as a result of re-duced population size. Such populations may then be-come more susceptible to mortality caused both by non-native pests and pathogens and by the environmentalpressures associated with climate change. This suscep-tibility could generate a cycle of mortality, loss of geneticvariation, and inability to adapt to change that could ul-timately result in population extirpation (Potter et al.,2010).

Along with the consideration of important species life-history traits and of threats other than climate change,such as pest and pathogen infestation (Dukes et al.,2009; Logan et al., 2003), we expect that the metricswe describe here will be valuable for scientists and pol-icymakers attempting to determine which forest treespecies, in the face of climate change, should be targetedfor monitoring efforts and for in situ and ex situ conser-vation actions such as seed banking efforts, facilitatedmigration, and genetic diversity studies. These measuresof predicted climate change pressure may be particularlyhelpful in multiple-species assessments across broad re-gional scales that take into account climate change riskto many species aggregated from relatively fine resolu-tion projection maps.

5 Acknowledgments

We thank Frank Koch for assistance with the ForestInventory and Analysis data, Steve Norman and Carol

Aubry for helpful suggestions, and Julie Canavin fordata organization. Forrest Hoffman was instrumental indeveloping the parallel MSTC code and ran the cluster-ing analysis. Barry Baker and Chris Zganjar preparedthe input data sets used for the MSTC process. Wethank the guest editors and the three anonymous review-ers for their helpful comments. This work was supportedin part by the Eastern Environmental Threat Assess-ment Center; by the Forest Health Monitoring programof the United States Department of Agriculture (USDA)Forest Service; and by Cooperative Agreement 09-CA-11330146-078 and Research Joint Venture Agreement10-JV-11330146-049 between the USDA Forest Service,Southern Research Station, and North Carolina StateUniversity.

References

Aitken, S.N., S. Yeaman, J.A. Holliday, T.L. Wang, andS. Curtis-McLane. 2008. Adaptation, migration or ex-tirpation: climate change outcomes for tree popula-tions. Evol. Appl. 1(1):95-111.

Aubry, C., W. Devine, R. Shoal, A. Bower, J. Miller, andN. Maggiulli. 2011. Climate Change and Forest Biodi-versity: A Vulnerability Assessment and Action Planfor National Forests in Western Washington. USDAFor. Serv. Pacific Northwest Region. 308 p.

Baker, B., H. Diaz, W. Hargrove, and F. Hoffman. 2010.Use of the Koppen-Trewartha climate classificationto evaluate climatic refugia in statistically derivedecoregions for the People’s Republic of China. Clim.Change 98(1-2):113-131.

Barazani, O., A. Perevolotsky, and R. Hadas. 2008. Aproblem of the rich: Prioritizing local plant geneticresources for ex situ conservation in Israel. Biol. Con-serv. 141(2):596-600.

Bechtold, W.A., and P.L. Patterson. 2005. The En-hanced Forest Inventory and Analysis Program: Na-tional Sampling Design and Estimation Procedures.USDA For. Serv. Gen. Tech. Rep. SRS-GTR-80. 85 p.

Brook, B.W., N.S. Sodhi, and C.J.A. Bradshaw. 2008.Synergies among extinction drivers under globalchange. Trends Ecol. Evol. 23(8):453-460.

Coates, D.J., and K.A. Atkins. 2001. Priority setting andthe conservation of Western Australia’s diverse andhighly endemic flora. Biol. Conserv. 97(2):251-263.

Crookston, N. 2013. Research on Forest ClimateChange: Potential Effects of Global Warming onForest and Plant Climate Relationships in West-ern North America and Mexico. Available online

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 161

at http://forest.moscowfsl.wsu.edu/climate/; last ac-cessed June 25, 2013.

Davis, M.B., and R.G. Shaw. 2001. Range shifts andadaptive responses to Quaternary climate change. Sci-ence 292(5517):673-679.

Davis, M.B., R.G. Shaw, and J.R. Etterson. 2005.Evolutionary responses to changing climate. Ecology86(7):1704-1714.

Devine, W., C. Aubry, J. Miller, K.M. Potter,and A.D. Bower. 2012. Climate change and foresttrees in the Pacific Northwest: Guide to vulner-ability assessment methodology. USDA For. Serv.Pacific Northwest Region. 49 p. Available onlineat http://ecoshare.info/projects/ccft/; last accessedAug. 15, 2013.

Dukes, J.S., J. Pontius, D. Orwig, J.R. Garnas, V.L.Rodgers, N. Brazee, B. Cooke, K.A. Theoharides, E.E.Stange, R. Harrington, J. Ehrenfeld, J. Gurevitch, M.Lerdau, K. Stinson, R. Wick, and M. Ayres. 2009. Re-sponses of insect pests, pathogens, and invasive plantspecies to climate change in the forests of northeast-ern North America: What can we predict? Can. J.For. Res. 39(2):231-248.

Fitzpatrick, M.C., and W.W. Hargrove. 2009. The pro-jection of species distribution models and the problemof non-analog climate. Biodivers. Conserv. 18(8):2255-2261.

Flora of North America Editorial Committee. 1993+.Flora of North America North of Mexico. Availableonline at http://floranorthamerica.org/; last accessedJune 25, 2013.

Fridley, J.D., D.B. Vandermast, D.M. Kuppinger, M.Manthey, and R.K. Peet. 2007. Co-occurrence basedassessment of habitat generalists and specialists: anew approach for the measurement of niche width. J.Ecol. 95(4):707-722.

Gauthier, P., M. Debussche, and J.D. Thompson. 2010.Regional priority setting for rare species based ona method combining three criteria. Biol. Conserv.143(6):1501-1509.

Global Soil Data Task Group. 2000. Global Gridded Sur-faces of Selected Soil Characteristics (InternationalGeosphere-Biosphere Programme - Data and Informa-tion System [IGBP-DIS]). Oak Ridge National Labo-ratory Distributed Active Archive Center, Oak Ridge,Tennessee.

Hampe, A., and R.J. Petit. 2005. Conserving biodiver-sity under climate change: the rear edge matters. Ecol.Lett. 8(5):461-467.

Hargrove, W.W., and F.M. Hoffman. 2003. An analyticalassessment tool for predicting changes in a species dis-tribution map following changes in environmental con-ditions. In Proceedings of the 4th International Con-ference on Integrating GIS and Environmental Mod-eling (GIS/EM4): Problems, Prospects and ResearchNeeds, September 2-8, 2000, Parks, B.O., K.M. Clark,and M.P. Crane (eds.). Cooperative Institute for Re-search in Environmental Sciences; NOAA NationalGeophysical Data Center, Ecosystem Informatics; andUS Geological Survey, Center for Biological Informat-ics, Banff, Alberta, Canada.

Hargrove, W.W., and F.M. Hoffman. 2005. Potentialof multivariate quantitative methods for delineationand visualization of ecoregions. Environ. Manage.34((Suppl. 1)):S39-S60.

Harris, J.A., R.J. Hobbs, E. Higgs, and J. Aron-son. 2006. Ecological restoration and global climatechange. Restor. Ecol. 14(2):170-176.

Hiebert, R.D., and J.L. Hamrick. 1984. An ecologicalstudy of bristleone pine (Pinus longaeva) in Utah andeastern Nevada. Great Basin Nat. 44(3):487-494.

Hijmans, R.J., S.E. Cameron, J.L. Parra, P.G. Jones,and A. Jarvis. 2005. Very high resolution interpolatedclimate surfaces for global land areas. Int. J. Climatol.25(15):1965-1978.

Hoffman, F.M., W.W. Hargrove, D.J. Erickson, and R.J.Oglesby. 2005. Using clustered climate regimes to an-alyze and compare predictions from fully coupled gen-eral circulation models. Earth Interact 9(10):1-27.

International Panel on Climate Change. 2007. Climatechange 2007: The physical science basis. Contributionof working group I to the fourth assessment report ofthe intergovernmental panel on climate change. Cam-bridge University Press, Cambridge, United Kingdomand New York, NY, USA. 996 p.

Iverson, L.R., and D. McKenzie. 2013. Tree-speciesrange shifts in a changing climate: detecting, mod-eling, assisting. Landscape Ecol. 28(5):879-889.

Iverson, L.R., A.M. Prasad, S.N. Matthews, and M. Pe-ters. 2008. Estimating potential habitat for 134 east-ern US tree species under six climate scenarios. For.Ecol. Manage. 254(3):390-406.

Iverson, L.R., M.W. Schwartz, and A.M. Prasad. 2004a.How fast and far might tree species migrate in theeastern United States due to climate change? Glob.Ecol. Biogeogr. 13(3):209-219.

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 162

Iverson, L.R., M.W. Schwartz, and A.M. Prasad. 2004b.Potential colonization of newly available tree-specieshabitat under climate change: an analysis for five east-ern US species. Landscape Ecol. 19(7):787-799.

Johnson, G.R., F.C. Sorensen, J.B. St Clair, and R.C.Cronn. 2004. Pacific Northwest forest tree seed zones:A template for native plants? Native Plants J.5(2):131-140.

Kowalski, C.J. 1972. On the effects of non-normality ondistribution of sample product-moment correlation co-efficient. J. R. Stat. Soc. Ser. C-Appl. Stat. 21(1):1-12.

Little, E.L. 1971. Atlas of United States Trees. Volume 1.Conifers and Important Hardwoods. USDA For. Serv.Misc. Pub. 1146. 9 p., illus. [313 maps, folio].

Logan, J.A., J. Regniere, and J.A. Powell. 2003. As-sessing the impacts of global warming on forest pestdynamics. Front. Ecol. Environ. 1(3):130-137.

Loo, J. 2009. Ecological impacts of non-indigenous inva-sive fungi as forest pathogens. Biol. Invasions 11(1):81-96.

Lugo, A.E., S.L. Brown, R. Dodson, T.S. Smith, andH.H. Shugart. 1999. The Holdridge life zones of theconterminous United States in relation to ecosystemmapping. J. Biogeogr. 26(5):1025-1038.

Malmsheimer, R.W., P. Heffernan, S. Brink, D. Cran-dall, F. Deneke, C. Galik, E. Gee, J.A. Helms, N.McClure, M. Mortimer, S. Ruddell, M. Smith, and J.Stewart. 2008. Forest management solutions for mit-igating climate change in the United States. J. For.106(3):115-173.

Mangold, R.D., and W.J. Libby. 1978. Model for re-forestation with optimal and suboptimal tree popu-lations. Silvae Genet. 27(2):66-68.

Matthews, S.N., L.R. Iverson, A.M. Prasad, M.P. Pe-ters, and P.G. Rodewald. 2011. Modifying climatechange habitat models using tree species-specific as-sessments of model uncertainty and life history-factors. For. Ecol. Manage. 262(8):1460-1472.

McLaughlin, S.P. 1995. An overview of the flora of thesky islands, southeastern Arizona: Diversity, affini-ties, and insularity. P. 60-70 in Biodiversity and Man-agement of the Madrean Archipelago: The Sky Is-lands of Southwestern United States and Northwest-ern Mexico, DeBano, L.H., P.H. Ffolliott, A. Ortega-Rubio, G.J. Gottfried, R.H. Hamre, and C.B. Edmin-ster (eds.). USDA For. Serv. Gen. Tech. Rep. RM-GTR-264. 669 p.

Moore, I.D., R.B. Grayson, and A.R. Ladson. 1991. Dig-ital terrain modeling: A review of hydrological, ge-omorphological, and biological applications. Hydrol.Process. 5(1):3-30.

Myking, T. 2002. Evaluating genetic resources of foresttrees by means of life history traits - a Norwegian ex-ample. Biodivers. Conserv. 11(9):1681-1696.

Neilson, R.P. 1995. A model for predicting continental-scale vegetation distribution and water balance. Ecol.Appl. 5(2):362-385.

Parmesan, C. 2006. Ecological and evolutionary re-sponses to recent climate change. Annu. Rev. Ecol.Evol. Syst. 37:637-669.

Parmesan, C., and G. Yohe. 2003. A globally coherentfingerprint of climate change impacts across naturalsystems. Nature 421(6918):37-42.

Pedlar, J.H., D.W. McKenney, I. Aubin, T. Beardmore,J. Beaulieu, L. Iverson, G.A. O’Neill, R.S. Winder,and C. Ste-Marie. 2012. Placing forestry in the as-sisted migration debate. Bioscience 62(9):835-842.

Petit, R.J., R. Bialozyt, P. Garnier-Gere, and A. Hampe.2004. Ecology and genetics of tree invasions: Fromrecent introductions to Quaternary migrations. For.Ecol. Manage. 197(1-3):117-137.

Potter, K.M., and W.W. Hargrove. 2012. Determin-ing suitable locations for seed transfer under climatechange: A global quantitative model. New Forest.43(5-6):581-599.

Potter, K.M., W.W. Hargrove, and F.H. Koch. 2010.Predicting climate change extirpation risk for cen-tral and southern Appalachian forest tree species. P.179-189 in Proceedings from the Conference on Ecol-ogy and Management of High-Elevation Forests of theCentral and Southern Appalachian Mountains, Snow-shoe, West Virginia, May 14-15, 2009, Rentch, J.S.,and T.M. Schuler (eds.). USDA For. Serv. Gen. Tech.Rep. NRS-P-64. 242 p.

Prasad, A.M., L.R. Iverson, S. Matthews,and M. Peters. 2013. A Climate ChangeAtlas for 134 Forest Tree Species of theEastern United States. Available online athttp://www.nrs.fs.fed.us/atlas/tree/tree atlas.html;last accessed June 25, 2013.

Prentice, I.C., W. Cramer, S.P. Harrison, R. Lee-mans, R.A. Monserud, and A.M. Solomon. 1992. Aglobal biome model based on plant physiology anddominance, soil properties and climate. J. Biogeogr.19(2):117-134.

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 163

Rehfeldt, G.E. 1999. Systematics and genetic structureof Ponderosae taxa (Pinaceae) inhabiting the moun-tain islands of the Southwest. Am. J. Bot. 86(5):741-752.

Rehfeldt, G.E., N.L. Crookston, M.V. Warwell, and J.S.Evans. 2006. Empirical analyses of plant-climate rela-tionships for the western United States. Int. J. PlantSci. 167(6):1123-1150.

Rehfeldt, G.E., C.C. Ying, D.L. Spittlehouse, and D.A.Hamilton. 1999. Genetic responses to climate in Pinuscontorta: Niche breadth, climate change, and refor-estation. Ecol. Monogr. 69(3):375-407.

Root, T.L., J.T. Price, K.R. Hall, S.H. Schneider, C.Rosenzweig, and J.A. Pounds. 2003. Fingerprints ofglobal warming on wild animals and plants. Nature421(6918):57-60.

Russell, E.W.B. 1987. Pre-blight distribution of Cas-tanea dentata (Marsh.) Borkh. Bull. Torrey Bot. Club114(2):183-190.

SAS Institute Inc. 2008. The SAS System for Windows,Version 9.2. Cary, North Carolina.

Saxon, E., B. Baker, W. Hargrove, F. Hoffman, and C.Zganjar. 2005. Mapping environments at risk underdifferent global climate change scenarios. Ecol. Lett.8(1):53-60.

Schwartz, M.W., L.R. Iverson, A.M. Prasad, S.N.Matthews, and R.J. O’Connor. 2006. Predicting ex-tinctions as a result of climate change. Ecology87(7):1611-1615.

Sjostrom, A., and C.L. Gross. 2006. Life-history charac-ters and phylogeny are correlated with extinction riskin the Australian angiosperms. J. Biogeogr. 33(2):271-290.

Smith, W.B. 2002. Forest inventory and analysis: a na-tional inventory and monitoring program. Environ.Pollut. 116:S233-S242.

St. Clair, J.B., and G.T. Howe. 2007. Genetic maladap-tation of coastal Douglas-fir seedlings to future cli-mates. Glob. Change Biol. 13(7):1441-1454.

Stenseth, N.C., A. Mysterud, G. Ottersen, J.W. Hurrell,K.S. Chan, and M. Lima. 2002. Ecological effects ofclimate fluctuations. Science 297(5585):1292-1296.

Stephenson, S.L., H.S. Adams, and M.L. Lipford. 1991.The present distribution of chestnut in the upland for-

est communities of Virginia. Bull. Torrey Bot. Club118(1):24-32.

Stork, N.E., J.A. Coddington, R.K. Colwell, R.L. Chaz-don, C.W. Dick, C.A. Peres, S. Sloan, and K. Willis.2009. Vulnerability and resilience of tropical forestspecies to land-use change. Conserv. Biol. 23(6):1438-1447.

Thomas, C.D., A. Cameron, R.E. Green, M. Bakkenes,L.J. Beaumont, Y.C. Collingham, B.F.N. Erasmus,M.F. de Siqueira, A. Grainger, L. Hannah, L. Hughes,B. Huntley, A.S. van Jaarsveld, G.F. Midgley, L.Miles, M.A. Ortega-Huerta, A.T. Peterson, O.L.Phillips, and S.E. Williams. 2004. Extinction risk fromclimate change. Nature 427(6970):145-148.

United States Geological Survey. 1999. Digital rep-resentation of ”Atlas of United States Trees”by Elbert L. Little Jr. Available online athttp://esp.cr.usgs.gov/data/atlas/little/; last ac-cessed June 25, 2013.

Vitt, P., K. Havens, A.T. Kramer, D. Sollenberger,and E. Yates. 2010. Assisted migration of plants:Changes in latitudes, changes in attitudes. Biol. Con-serv. 143(1):18-27.

Warshall, P. 1995. The Madrean sky island archipelago:A planetary overview. P. 6-18 in Biodiversity andManagement of the Madrean Archipelago: The SkyIslands of Southwestern United States and Northwest-ern Mexico, DeBano, L.H., P.H. Ffolliott, A. Ortega-Rubio, G.J. Gottfried, R.H. Hamre, and C.B. Edmin-ster (eds.). USDA For. Serv. Gen. Tech. Rep. RM-GTR-264. 669 p.

Woodall, C.W., C.M. Oswalt, J.A. Westfall, C.H. Perry,M.D. Nelson, and A.O. Finley. 2009. An indicator oftree migration in forests of the eastern United States.For. Ecol. Manage. 257(5):1434-1444.

Woudenberg, S.W., B.L. Conkling, B.M. O’Connell,E.B. LaPoint, J.A. Turner, and K.L. Waddell.2010. The Forest Inventory and Analysis Database:Database Description and Users Manual Version 4.0for Phase 2. USDA For. Serv. Gen. Tech. Rep. RMRS-GTR-245. 336 p.

Wu, H.X., and C.C. Ying. 2004. Geographic pattern oflocal optimality in natural populations of lodgepolepine. For. Ecol. Manage. 194(1-3):177-198.

Zhu, K., C.W. Woodall, and J.S. Clark. 2012. Failure tomigrate: Lack of tree range expansion in response toclimate change. Glob. Change Biol. 18(3):1042-1052.

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 164

Appendix

A

Tab

le4:

Stu

dy

spec

ies,

incl

ud

ing

regi

onas

sign

men

t,tr

eesi

zecl

ass

,nu

mb

erof

Fore

stIn

vento

ryan

dA

naly

sis

(FIA

)p

lots

,an

dcl

imate

chan

ge

pre

ssu

rem

etri

csd

evel

oped

usi

ng

the

Mu

ltiv

aria

teS

pat

io-T

emp

oral

Clu

ster

ing

ap

pro

ach

(Su

itab

leH

ab

itat

Are

a,

Ran

ge

Sta

bil

ity,

Sh

ift

Pre

ssure

,an

dN

ich

eO

ccu

pan

cy).

FIA

Suitable

HabitatAre

a(k

m2)

Range

Sta

bility

Shift

Pre

ssure

Niche

Occ.

SpeciesName

Common

Name

Region

Size

Plots

Current

Futu

re%

Chg.

Are

a(k

m2)

%M

ean

(cells)

SD

Abie

sgra

ndis

gran

dfi

rW

11,2

19

561,5

62

452,9

56

-19.3

4362,5

73

64.5

73.6

19.2

2136

Abie

spro

cera

nob

lefi

rW

1154

81,8

64

60,4

36

-26.1

853,9

62

65.9

24.1

08.9

521

Ace

rba

rbatu

mF

lori

da

map

leE

S1

454

746,2

24

237,5

95

-68.1

6118,4

32

15.8

712.1

017.2

169

Ace

rgla

bru

mR

ock

yM

ounta

inm

aple

W2

208

163,9

15

112,9

90

-31.0

783,2

44

50.7

83.2

514.8

736

Ace

rgra

ndid

en

tatu

mb

igto

oth

map

leW

1121

104,2

41

76,8

14

-26.3

142,1

11

40.4

05.4

412.1

024

Ace

rle

uco

derm

ech

alk

map

leE

S1

21

116,8

97

8,6

61

-92.5

969

0.0

639.3

430.4

96

Ace

rm

acro

phyll

um

big

leaf

map

leW

1679

323,5

87

267,8

37

-17.2

3219,8

02

67.9

35.4

826.6

186

Ace

rn

egu

ndo

box

eld

erE

G1

2,4

55

2,4

46,2

35

1,0

52,7

62

-56.9

6834,0

34

34.0

96.9

411.8

3276

Ace

rn

igru

mb

lack

map

leE

N1

127

379,7

79

50,4

95

-86.7

022,5

45

5.9

419.3

823.8

728

Ace

rpen

sylv

an

icu

mst

rip

edm

aple

EN

2562

602,8

25

354,6

67

-41.1

7210,1

79

34.8

715.8

244.1

296

Ace

rru

bru

mre

dm

aple

EG

129,5

35

3,6

18,1

16

2,8

53,8

68

-21.1

22,2

51,5

43

62.2

35.3

513.6

9635

Ace

rsa

cchari

nu

msi

lver

map

leE

G1

1,1

34

1,5

58,1

65

461,9

20

-70.3

5364,8

04

23.4

112.9

521.9

9168

Ace

rsa

ccharu

msu

gar

map

leE

N1

14,0

41

2,9

07,7

38

1,8

38,0

90

-36.7

91,3

77,1

69

47.3

67.5

314.6

7473

Aesc

ulu

sfl

ava

yell

owb

uck

eye

EG

1378

436,9

87

165,6

82

-62.0

9100,2

84

22.9

522.1

642.5

452

Aesc

ulu

sgla

bra

Oh

iob

uck

eye

EN

1288

626,7

07

135,9

43

-78.3

191,1

14

14.5

418.2

326.0

654

Aln

us

rubra

red

ald

erW

11,0

01

291,3

95

233,7

73

-19.7

7186,0

94

63.8

69.5

370.2

079

Am

ela

nchie

rsp

p.

com

mon

serv

iceb

erry

EG

22,0

61

1,8

06,6

84

773,8

25

-57.1

7572,6

02

31.6

99.9

214.2

4237

Arb

utu

sm

en

ziesi

iP

acifi

cm

adro

ne

W2

716

232,9

00

197,0

66

-15.3

9141,7

40

60.8

63.9

711.5

665

Asi

min

atr

iloba

paw

paw

EG

3215

551,8

87

136,4

75

-75.2

739,6

25

7.1

817.7

819.4

647

Betu

laall

eghan

ien

sis

yell

owb

irch

EN

15,5

39

1,5

38,5

73

1,1

21,2

60

-27.1

2729,0

94

47.3

99.6

521.2

5290

Betu

lale

nta

swee

tb

irch

EG

12,6

55

946,1

12

623,7

69

-34.0

7405,1

31

42.8

224.5

742.6

5144

Betu

lan

igra

rive

rbir

chE

G1

706

1,2

51,2

36

451,6

42

-63.9

0298,8

93

23.8

96.1

410.1

6120

Betu

lapopu

lifo

lia

gray

bir

chE

N1

488

443,0

36

345,4

80

-22.0

2217,0

79

49.0

08.6

116.8

072

Calo

cedru

sdec

urr

en

sin

cen

se-c

edar

W1

989

330,4

25

268,1

84

-18.8

4177,2

78

53.6

53.1

58.1

986

Carp

inu

sca

roli

nia

na

mu

scle

wood

EG

22,8

53

1,8

54,2

68

1,0

25,5

54

-44.6

9872,4

69

47.0

54.3

18.9

0218

Cary

aalb

am

ock

ernut

hic

kory

EG

16,5

91

2,2

05,0

14

1,3

42,0

74

-39.1

41,2

00,1

34

54.4

33.2

26.7

6292

Cary

aaqu

ati

caw

ater

hic

kory

ES

1455

575,4

03

300,7

01

-47.7

4157,0

33

27.2

97.6

112.5

866

Cary

aco

rdif

orm

isb

itte

rnu

th

icko

ryE

G1

2,7

98

2,2

38,2

60

795,6

41

-64.4

5661,3

85

29.5

55.5

09.7

5252

Continued

onnextpage

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 165

Table

4–

con

tin

ued

from

pre

vio

us

page

FIA

Suitable

HabitatAre

a(k

m2)

Range

Sta

bility

Shift

Pre

ssure

Niche

Occ.

SpeciesName

Common

Name

Region

Size

Plots

Current

Futu

re%

Chg.

Are

a(k

m2)

%M

ean

(cells)

SD

Cary

agla

bra

pig

nu

th

icko

ryE

G1

6,4

93

2,1

16,6

78

1,1

97,9

64

-43.4

01,0

11,1

58

47.7

75.6

89.7

0273

Cary

ail

lin

oin

en

sis

pec

anE

G1

483

817,5

61

281,2

09

-65.6

0135,2

44

16.5

410.0

017.6

586

Cary

ala

cin

iosa

shel

lbar

kh

ickor

yE

N1

347

672,5

21

173,0

63

-74.2

760,4

04

8.9

813.9

417.2

660

Cary

aovata

shag

bar

kh

ickor

yE

G1

4,4

41

2,2

90,0

60

1,0

46,5

08

-54.3

0898,3

19

39.2

35.1

48.1

2281

Cary

apall

ida

san

dh

icko

ryE

S1

92

289,2

97

76,2

58

-73.6

419,3

90

6.7

025.8

730.4

821

Cary

ate

xan

ab

lack

hic

kory

ES

12,2

75

969,5

60

603,3

88

-37.7

7317,4

57

32.7

45.0

16.4

9128

Cast

an

eaden

tata

Am

eric

anch

estn

ut

EG

3121

228,4

70

77,9

77

-65.8

739,9

85

17.5

028.4

065.4

627

Celt

isla

evig

ata

suga

rber

ryE

S1

1,5

01

1,2

89,4

09

689,8

98

-46.5

0476,8

20

36.9

84.1

78.3

6153

Celt

isocc

iden

tali

sh

ackb

erry

EG

12,9

40

2,3

63,6

56

838,5

48

-64.5

2711,2

93

30.0

95.6

18.8

9260

Cerc

isca

naden

sis

east

ern

red

bu

dE

G2

1,7

96

1,6

11,8

26

629,6

46

-60.9

4427,1

41

26.5

09.3

915.5

4178

Cham

aec

ypari

sth

y-

oid

es

Atl

anti

cw

hit

eced

arE

G1

67

115,5

04

65,5

41

-43.2

646,3

90

40.1

619.9

037.8

714

Chry

sole

pis

chry

so-

phyll

ago

lden

chin

qu

apin

W2

200

156,8

54

109,8

41

-29.9

772,5

44

46.2

55.0

041.0

138

Corn

us

flori

da

flow

erin

gd

ogw

ood

EG

24,3

01

1,7

42,2

96

935,8

84

-46.2

8797,7

91

45.7

93.4

86.5

1217

Cra

taeg

ussp

p.

haw

thor

nsp

ecie

sE

G2

1,6

67

2,0

83,6

60

749,2

07

-64.0

4554,6

45

26.6

26.1

612.4

1223

Dio

spyro

svir

gin

ian

aco

mm

onp

ersi

mm

onE

G1

1,6

36

1,4

90,2

20

766,2

78

-48.5

8583,8

57

39.1

84.1

87.4

2177

Fagu

sgra

ndif

oli

aA

mer

ican

bee

chE

G1

6,7

95

2,4

11,9

95

1,5

31,6

51

-36.5

01,1

72,6

98

48.6

24.3

811.2

2364

Fra

xin

us

am

eri

can

aw

hit

eas

hE

G1

8,3

15

2,8

93,7

10

1,7

35,3

12

-40.0

31,4

29,9

39

49.4

23.5

96.8

1438

Fra

xin

us

caro

lin

ian

aC

arol

ina

ash

ES

177

163,6

37

105,0

27

-35.8

275,6

83

46.2

512.9

215.0

916

Fra

xin

us

lati

foli

aO

rego

nas

hW

148

27,5

04

23,3

60

-15.0

78,7

24

31.7

215.9

674.8

46

Fra

xin

us

nig

rab

lack

ash

EN

13,5

44

1,1

18,3

21

661,9

39

-40.8

1394,9

97

35.3

218.0

130.0

6183

Fra

xin

us

pen

nsy

lvan

-ic

agr

een

ash

EG

16,9

50

3,7

83,7

19

2,0

74,4

19

-45.1

81,7

15,2

58

45.3

35.6

010.7

8518

Fra

xin

us

pro

fun

da

pu

mp

kin

ash

ES

150

94,2

45

58,9

34

-37.4

748,2

27

51.1

724.4

141.9

69

Fra

xin

us

qu

adra

ngu

-la

tab

lue

ash

EN

1149

310,9

97

74,2

22

-76.1

319,4

22

6.2

534.5

834.7

325

Fra

xin

us

texen

sis

Tex

asas

hE

S1

26

63,9

77

5,3

51

-91.6

44,6

88

7.3

311.5

911.9

15

Gle

dit

sia

aqu

ati

caw

ater

locu

stE

S1

72

125,4

23

46,5

03

-62.9

214,8

30

11.8

234.6

037.0

213

Gle

dit

sia

tria

can

thos

hon

eylo

cust

EG

11,2

09

1,6

90,4

49

549,7

84

-67.4

8429,7

77

25.4

26.2

711.9

3174

Gord

on

iala

sian

thu

slo

blo

lly

bay

ES

1276

219,0

36

180,1

35

-17.7

6148,0

20

67.5

88.8

613.1

028

Gym

nocla

du

sdio

icu

sK

entu

cky

coff

eetr

eeE

N1

67

149,3

96

46,8

48

-68.6

41,8

89

1.2

654.3

532.2

717

Hale

sia

caro

lin

ian

aC

arol

ina

silv

erb

ell

ES

322

126,0

07

43,1

34

-65.7

722,0

96

17.5

417.0

712.9

214

Continued

onnextpage

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 166

Table

4–

con

tin

ued

from

pre

vio

us

page

FIA

Suitable

HabitatAre

a(k

m2)

Range

Sta

bility

Shift

Pre

ssure

Niche

Occ.

SpeciesName

Common

Name

Region

Size

Plots

Current

Futu

re%

Chg.

Are

a(k

m2)

%M

ean

(cells)

SD

Ilex

opaca

Am

eric

anh

olly

ES

22,5

16

1,2

57,4

75

824,5

53

-34.4

3686,4

23

54.5

93.6

911.2

3166

Ju

gla

ns

cin

ere

ab

utt

ernu

tE

N1

289

702,4

05

131,5

06

-81.2

877,6

55

11.0

620.8

934.7

856

Ju

gla

ns

nig

rab

lack

wal

nu

tE

G1

3,5

89

2,1

12,4

99

717,4

23

-66.0

4613,9

00

29.0

610.9

318.1

5229

Ju

nip

eru

sm

on

osp

erm

aon

esee

dju

nip

erW

21,5

43

1,0

53,2

23

723,7

83

-31.2

8546,9

05

51.9

34.5

912.5

9168

Ju

nip

eru

sos-

teosp

erm

aU

tah

jun

iper

W1

3,3

19

1,5

95,1

35

1,1

95,0

58

-25.0

8917,2

77

57.5

04.9

919.8

2285

Ju

nip

eru

ssc

opu

lo-

rum

Rock

yM

ounta

inju

-n

iper

W2

1,5

85

1,3

66,9

43

740,5

90

-45.8

2619,3

12

45.3

14.0

89.0

4221

Ju

nip

eru

svir

gin

ian

aea

ster

nre

dce

dar

EG

15,5

75

2,6

80,5

55

1,2

65,5

99

-52.7

91,0

62,5

20

39.6

45.0

09.7

0332

Lari

xocc

iden

tali

sw

este

rnla

rch

W1

962

426,8

00

359,3

98

-15.7

9267,4

60

62.6

73.0

68.4

3101

Liq

uid

am

bar

styra

ci-

flu

asw

eetg

um

ES

113,8

22

1,8

08,6

86

1,4

47,1

17

-19.9

91,1

60,5

54

64.1

73.3

710.8

5268

Lir

ioden

dro

ntu

lip-

ifera

yell

ow-p

opla

rE

G1

9,8

02

1,7

85,1

69

1,1

62,7

44

-34.8

7999,9

65

56.0

24.0

09.0

4246

Magn

oli

aacu

min

ata

cucu

mb

ertr

eeE

G1

697

767,3

05

286,0

05

-62.7

3226,4

73

29.5

216.7

524.2

083

Magn

oli

afr

ase

rim

ounta

inm

agnol

iaE

S1

216

213,9

60

108,6

54

-49.2

255,9

11

26.1

321.4

547.2

935

Magn

oli

agra

ndifl

ora

Sou

ther

nm

agn

olia

ES

1407

466,5

63

376,0

23

-19.4

1182,4

08

39.1

06.4

912.7

956

Magn

oli

am

acro

-phyll

ab

igle

afm

agn

olia

ES

1139

224,3

20

96,2

59

-57.0

928,5

15

12.7

19.5

212.2

024

Magn

oli

atr

ipeta

lau

mb

rell

am

agn

olia

EG

245

130,3

64

31,4

61

-75.8

77,4

48

5.7

134.2

458.6

212

Magn

oli

avir

gin

ian

asw

eetb

ayE

S1

2,1

15

925,5

10

677,8

88

-26.7

6509,7

80

55.0

86.0

410.7

1119

Malu

san

gu

stif

oli

aso

uth

ern

crab

app

leE

S2

48

154,9

33

26,5

13

-82.8

916,4

07

10.5

922.1

518.7

39

Malu

sco

ron

ari

asw

eet

crab

app

leE

N2

40

107,5

23

7,7

00

-92.8

42,2

64

2.1

116.2

520.2

27

Moru

sru

bra

red

mu

lber

ryE

G1

1,4

14

1,9

45,2

83

609,4

75

-68.6

7449,3

58

23.1

06.2

110.3

6197

Nyss

aaqu

ati

caw

ater

tup

elo

ES

1464

743,4

90

337,1

66

-54.6

5213,5

85

28.7

34.6

910.1

177

Nyss

abifl

ora

swam

ptu

pel

oE

S1

2,1

42

930,4

00

567,8

36

-38.9

7461,6

06

49.6

14.3

212.2

7115

Nyss

aogec

he

Oge

ech

eetu

pel

oE

S1

49

79,4

55

30,4

14

-61.7

225,1

83

31.6

93.5

95.8

18

Nyss

asy

lvati

cab

lack

gum

EG

17,8

98

2,1

17,4

67

1,4

81,5

61

-30.0

31,2

48,7

16

58.9

73.4

47.1

3304

Ost

rya

vir

gin

ian

aea

ster

nh

oph

ornb

eam

EG

24,3

77

2,9

28,9

95

1,5

05,4

22

-48.6

01,1

70,2

52

39.9

55.0

08.6

2392

Oxyden

dru

mar-

bore

um

sou

rwood

EG

13,9

48

1,2

36,3

47

661,3

96

-46.5

0538,7

77

43.5

84.5

17.7

0153

Pers

eabo

rbon

iare

db

ayE

S2

759

544,7

66

456,6

60

-16.1

7304,6

82

55.9

38.6

117.5

776

Continued

onnextpage

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 167

Table

4–

con

tin

ued

from

pre

vio

us

page

FIA

Suitable

HabitatAre

a(k

m2)

Range

Sta

bility

Shift

Pre

ssure

Niche

Occ.

SpeciesName

Common

Name

Region

Size

Plots

Current

Futu

re%

Chg.

Are

a(k

m2)

%M

ean

(cells)

SD

Pic

earu

ben

sre

dsp

ruce

EN

12,2

74

554,8

07

514,9

99

-7.1

8295,6

94

53.3

06.5

39.6

2115

Pic

easi

tchen

sis

Sit

kasp

ruce

W1

667

180,6

97

166,0

53

-8.1

0125,1

93

69.2

88.3

340.5

055

Pin

us

ari

stata

Rock

yM

ounta

inb

rist

le-

con

ep

ine

W2

85

122,1

84

60,9

64

-50.1

051,8

36

42.4

24.9

017.0

020

Pin

us

att

en

uata

Kn

obco

ne

pin

eW

168

56,5

94

48,6

03

-14.1

218,3

05

32.3

48.5

961.6

816

Pin

us

balf

ou

rian

afo

xta

ilp

ine

W1

19

13,5

03

5,1

86

-61.5

92,3

24

17.2

128.0

456.6

53

Pin

us

echin

ata

shor

tlea

fp

ine

ES

15,2

24

1,4

40,9

49

949,7

28

-34.0

9723,5

87

50.2

23.6

510.8

4197

Pin

us

flexil

isli

mb

erp

ine

W1

573

529,4

26

272,8

64

-48.4

6230,6

70

43.5

78.2

923.4

788

Pin

us

gla

bra

spru

cep

ine

ES

1252

345,2

37

199,6

12

-42.1

895,5

53

27.6

83.5

44.2

839

Pin

us

jeff

reyi

Jeff

rey

pin

eW

1635

331,2

98

216,9

70

-34.5

1145,9

10

44.0

43.6

812.9

171

Pin

us

lam

bert

ian

asu

gar

pin

eW

1852

276,8

72

215,7

43

-22.0

8141,3

97

51.0

73.2

69.4

471

Pin

us

lon

gaeva

Gre

atB

asin

bri

stle

con

ep

ine

W2

19

2,8

82

5,8

93

104.4

8726

25.1

973.3

4121.1

91

Pin

us

mon

tico

law

este

rnw

hit

ep

ine

W1

499

303,6

42

215,2

01

-29.1

3167,4

06

55.1

33.5

811.8

071

Pin

us

palu

stri

slo

ngl

eaf

pin

eE

S1

1,5

25

753,7

76

622,2

82

-17.4

4467,2

59

61.9

96.7

113.1

8103

Pin

us

pon

dero

sap

ond

eros

ap

ine

W1

5,6

70

2,1

93,0

29

1,5

15,0

92

-30.9

11,2

33,6

61

56.2

56.0

114.0

1435

Pin

us

pu

ngen

sT

able

Mou

nta

inp

ine

EG

182

171,1

26

45,0

78

-73.6

626,0

20

15.2

132.5

766.0

219

Pin

us

resi

nosa

red

pin

eE

N1

2,3

98

909,9

01

551,0

59

-39.4

4322,4

08

35.4

313.6

424.8

4151

Pin

us

rigid

ap

itch

pin

eE

G1

626

669,6

03

316,7

96

-52.6

9199,5

94

29.8

121.9

039.6

990

Pin

us

sabin

ian

agr

ayp

ine

W1

223

133,3

37

111,6

01

-16.3

059,0

66

44.3

05.6

819.5

735

Pin

us

sero

tin

ap

ond

pin

eE

S1

320

309,8

39

207,1

75

-33.1

3165,4

29

53.3

96.9

926.1

035

Pin

us

stro

bu

sw

hit

ep

ine

EG

12,5

01

1,8

96,5

56

1,2

54,5

57

-33.8

5854,3

20

45.0

59.0

117.3

0316

Pin

us

taed

alo

blo

lly

pin

eE

S1

14,2

23

1,5

11,8

03

1,1

63,3

62

-23.0

5980,1

46

64.8

35.1

913.9

0221

Pin

us

vir

gin

ian

aV

irgi

nia

pin

eE

G1

2,5

58

974,6

82

444,7

50

-54.3

7329,0

04

33.7

66.4

110.1

0120

Pla

tan

us

occ

iden

tali

sA

mer

ican

syca

mor

eE

G1

2,0

14

1,8

38,8

58

803,6

41

-56.3

0641,8

52

34.9

04.2

89.3

2205

Popu

lus

delt

oid

es

east

ern

cott

onw

ood

EG

1851

1,4

83,9

34

435,1

54

-70.6

8302,3

36

20.3

712.6

820.6

4151

Popu

lus

gra

ndid

en

-ta

tab

igto

oth

asp

enE

N1

3,5

04

1,4

27,6

66

900,4

32

-36.9

3614,1

41

43.0

210.4

823.1

3228

Pru

nu

sam

eri

can

aA

mer

ican

plu

mE

G3

200

576,6

66

119,1

11

-79.3

442,3

85

7.3

521.8

623.7

545

Pru

nu

sse

roti

na

bla

ckch

erry

EG

112,1

04

3,3

02,8

52

2,1

51,2

39

-34.8

71,7

87,8

25

54.1

33.9

37.6

5502

Pse

udots

uga

men

-zi

esi

iD

ougl

as-fi

rW

18,5

36

1,8

25,0

92

1,3

47,3

66

-26.1

81,1

50,6

96

63.0

57.1

118.6

7429

Qu

erc

us

agri

foli

aC

alif

orn

iali

veoa

kW

1186

124,8

92

82,0

39

-34.3

135,6

02

28.5

17.0

312.3

126

Continued

onnextpage

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 168

Table

4–

con

tin

ued

from

pre

vio

us

page

FIA

Suitable

HabitatAre

a(k

m2)

Range

Sta

bility

Shift

Pre

ssure

Niche

Occ.

SpeciesName

Common

Name

Region

Size

Plots

Current

Futu

re%

Chg.

Are

a(k

m2)

%M

ean

(cells)

SD

Qu

erc

us

alb

aw

hit

eoa

kE

G1

15,8

24

2,8

52,2

92

2,0

00,0

38

-29.8

81,6

89,4

19

59.2

33.1

86.3

7434

Qu

erc

us

ari

zon

ica

Ari

zon

aw

hit

eoa

kW

1348

237,2

15

168,3

18

-29.0

497,7

30

41.2

011.6

226.4

550

Qu

erc

us

bic

olo

rsw

amp

wh

ite

oak

EN

1392

621,2

60

188,5

60

-69.6

5123,3

84

19.8

610.2

220.2

469

Qu

erc

us

chry

sole

pis

canyo

nli

veoa

kW

1692

272,7

58

214,0

14

-21.5

4145,1

50

53.2

23.6

98.2

372

Qu

erc

us

cocc

inea

scar

let

oak

EN

14,3

98

1,4

85,3

70

768,4

03

-48.2

7557,3

17

37.5

27.2

111.7

5195

Qu

erc

us

dou

gla

sii

blu

eoa

kW

1327

191,6

13

187,2

90

-2.2

687,1

37

45.4

84.8

911.2

750

Qu

erc

us

ell

ipso

idali

sn

orth

ern

pin

oak

EN

11,1

46

602,7

13

267,8

78

-55.5

5189,9

96

31.5

215.8

220.4

485

Qu

erc

us

em

ory

iE

mor

yoa

kW

2226

166,4

41

90,7

48

-45.4

846,5

99

28.0

010.5

036.0

932

Qu

erc

us

falc

ata

sou

ther

nre

doa

kE

S1

5,5

16

1,6

06,8

04

1,2

03,4

92

-25.1

0935,7

69

58.2

43.3

36.1

3228

Qu

erc

us

gam

beli

iG

amb

eloa

kW

31,7

96

1,0

59,0

01

680,1

21

-35.7

8533,7

44

50.4

06.2

214.7

2194

Qu

erc

us

garr

yan

aO

rego

nw

hit

eoa

kW

3313

269,2

40

216,1

82

-19.7

1133,8

40

49.7

15.0

211.5

669

Qu

erc

us

gri

sea

gray

oak

W3

19

33,6

50

15,9

45

-52.6

25,3

87

16.0

141.2

640.7

04

Qu

erc

us

ilic

ifoli

asc

rub

oak

EN

352

121,4

93

55,6

53

-54.1

910,5

06

8.6

526.4

368.4

812

Qu

erc

us

imbri

cari

ash

ingl

eoa

kE

N1

606

767,1

97

216,4

21

-71.7

9120,7

11

15.7

318.8

723.2

876

Qu

erc

us

inca

na

blu

ejac

koa

kE

S2

194

351,7

06

145,2

55

-58.7

089,8

69

25.5

514.8

818.3

534

Qu

erc

us

kell

oggii

Cal

ifor

nia

bla

ckoa

kW

1863

247,8

27

218,4

51

-11.8

5144,7

39

58.4

03.6

25.8

971

Qu

erc

us

laevis

turk

eyoa

kE

S2

336

365,6

06

234,4

37

-35.8

8185,4

15

50.7

16.4

012.1

543

Qu

erc

us

lau

rifo

lia

lau

rel

oak

ES

12,3

27

946,8

61

689,9

00

-27.1

4524,1

84

55.3

66.9

016.1

5128

Qu

erc

us

loba

taC

alif

orn

iaw

hit

eoa

kW

170

59,0

50

47,8

87

-18.9

016,2

33

27.4

95.8

424.4

915

Qu

erc

us

lyra

taov

ercu

poa

kE

S1

660

754,0

90

425,0

91

-43.6

3257,1

53

34.1

03.4

97.0

688

Qu

erc

us

macro

carp

ab

ur

oak

EN

12,2

50

1,4

99,0

70

528,8

39

-64.7

2394,3

86

26.3

111.3

019.3

0183

Qu

erc

us

marg

are

ttia

ed

war

fp

ost

oak

ES

2175

230,2

03

96,9

53

-57.8

867,0

67

29.1

37.2

823.7

522

Qu

erc

us

mari

lan

dic

ab

lack

jack

oak

ES

11,2

16

1,1

59,3

98

526,7

28

-54.5

7295,3

36

25.4

74.5

05.4

5126

Qu

erc

us

mic

hau

xii

swam

pch

estn

ut

oak

ES

1624

961,2

33

510,7

81

-46.8

6385,5

02

40.1

04.3

98.9

298

Qu

erc

us

min

ima

dw

arf

live

oak

ES

2102

106,9

51

123,7

42

15.7

090,7

12

84.8

211.8

456.5

319

Qu

erc

us

mu

ehle

n-

berg

iich

inka

pin

oak

EG

11,5

48

1,2

14,1

97

498,1

26

-58.9

7280,0

74

23.0

713.0

617.1

9139

Qu

erc

us

nig

raw

ater

oak

ES

17,5

86

1,3

80,4

79

1,1

52,3

78

-16.5

2861,9

04

62.4

43.2

410.4

6207

Qu

erc

us

pagoda

cher

ryb

ark

oak

ES

11,7

06

1,1

44,0

16

654,2

50

-42.8

1497,9

31

43.5

23.0

57.3

5135

Qu

erc

us

palu

stri

sp

inoa

kE

N1

493

805,6

17

206,1

99

-74.4

0119,4

71

14.8

312.9

220.4

381

Qu

erc

us

phell

os

wil

low

oak

ES

11,6

92

1,2

08,2

21

693,4

67

-42.6

0560,5

76

46.4

04.0

66.7

9142

Qu

erc

us

pri

nu

sch

estn

ut

oak

EG

14,6

33

1,1

52,1

98

646,0

68

-43.9

3458,8

12

39.8

26.3

310.6

1161

Qu

erc

us

rubra

nor

ther

nre

doa

kE

G1

12,2

32

2,9

13,0

20

1,8

98,4

49

-34.8

31,4

75,5

75

50.6

55.9

312.0

0456

Continued

onnextpage

Potter and Hargrove (2013)/Math.Comput. For.Nat.-Res. Sci. Vol. 5, Issue 2, pp. 151–169/http://mcfns.com 169

Table

4–

con

tin

ued

from

pre

vio

us

page

FIA

Suitable

HabitatAre

a(k

m2)

Range

Sta

bility

Shift

Pre

ssure

Niche

Occ.

SpeciesName

Common

Name

Region

Size

Plots

Current

Futu

re%

Chg.

Are

a(k

m2)

%M

ean

(cells)

SD

Qu

erc

us

shu

mard

iiS

hu

mar

doa

kE

S1

442

886,9

56

298,5

12

-66.3

4150,5

34

16.9

79.8

116.8

791

Qu

erc

us

sim

ilis

Del

tap

ost

oak

ES

125

20,9

10

1,7

42

-91.6

7131

0.6

361.2

831.0

02

Qu

erc

us

sin

uata

Du

ran

doa

kE

S1

31

45,4

85

8,4

62

-81.4

02,7

33

6.0

129.4

117.6

74

Qu

erc

us

stell

ata

pos

toa

kE

S1

6,7

91

1,9

58,0

86

1,4

60,4

39

-25.4

11,0

37,5

26

52.9

92.8

25.9

0281

Qu

erc

us

texan

aN

uta

lloa

kE

S1

270

292,5

39

181,8

96

-37.8

299,5

53

34.0

34.1

313.0

538

Qu

erc

us

velu

tin

ab

lack

oak

EG

18,8

07

2,5

33,6

33

1,5

74,9

52

-37.8

41,2

96,8

40

51.1

84.1

36.9

3360

Qu

erc

us

vir

gin

ian

ali

veoa

kE

S1

1,1

93

739,9

00

493,2

94

-33.3

3376,9

68

50.9

57.6

322.4

5106

Qu

erc

us

wis

lize

ni

inte

rior

live

oak

W1

253

170,0

49

139,8

51

-17.7

669,2

17

40.7

04.8

810.5

843

Robin

iapse

udoaca

cia

bla

cklo

cust

EG

12,1

60

1,5

67,9

69

625,6

49

-60.1

0510,2

27

32.5

49.5

919.0

4184

Sali

xn

igra

bla

ckw

illo

wE

G2

1,1

60

2,0

12,8

51

687,6

06

-65.8

4521,8

29

25.9

29.6

118.4

8201

Sass

afr

as

alb

idu

msa

ssaf

ras

EG

13,7

60

1,9

96,1

48

1,0

45,7

36

-47.6

1833,9

98

41.7

83.2

25.3

5259

Seq

uoia

sem

perv

iren

sco

ast

red

wood

W1

202

95,7

12

78,4

85

-18.0

053,9

58

56.3

88.5

431.2

523

Sid

ero

xylo

nla

nu

gi-

nosu

mgu

mb

um

elia

ES

2183

372,5

24

82,2

75

-77.9

125,8

26

6.9

312.9

924.5

637

Taxodiu

mdis

tichu

mb

ald

cyp

ress

ES

1894

847,3

44

550,6

03

-35.0

2382,5

09

45.1

46.5

813.6

3108

Taxu

sbre

vif

oli

aP

acifi

cye

wW

2129

130,0

26

75,4

33

-41.9

957,1

15

43.9

34.1

735.7

528

Thu

jaocc

iden

tali

sn

orth

ern

wh

ite-

ced

arE

N1

4,2

78

1,0

35,2

30

808,4

47

-21.9

1516,5

65

49.9

010.7

340.8

7206

Til

iaam

eri

can

aA

mer

ican

bas

swood

EN

15,0

02

2,4

42,8

40

1,0

69,0

28

-56.2

4767,8

96

31.4

312.1

019.3

5326

Til

iaam

eri

can

avar.

caro

lin

ian

aC

arol

ina

bas

swood

ES

147

52,2

63

51,5

42

-1.3

810,1

67

19.4

514.2

215.0

28

Til

iaam

eri

can

avar.

hete

rophyll

aw

hit

eb

assw

ood

EG

145

67,1

96

17,2

11

-74.3

95,6

77

8.4

533.0

333.9

97

Tsu

ga

can

aden

sis

east

ern

hem

lock

EG

14,4

85

1,5

37,1

30

1,0

35,4

42

-32.6

4689,9

87

44.8

913.8

032.1

3261

Tsu

ga

caro

lin

ian

aC

arol

ina

hem

lock

ES

119

44,2

17

29,5

75

-33.1

115,9

45

36.0

649.4

381.2

39

Tsu

ga

mert

en

sian

am

ounta

inh

emlo

ckW

1828

311,0

59

220,1

04

-29.2

4175,5

89

56.4

54.6

110.7

480

Ulm

us

ala

taw

inge

del

mE

S1

5,4

13

1,4

98,9

70

1,0

30,1

04

-31.2

8705,7

46

47.0

82.5

23.9

3199

Ulm

us

am

eri

can

aA

mer

ican

elm

EG

19,5

20

3,6

57,4

37

2,1

35,6

89

-41.6

11,7

44,2

12

47.6

96.2

513.0

9522

Ulm

us

cra

ssif

oli

ace

dar

elm

ES

1414

498,4

12

217,2

52

-56.4

1101,3

21

20.3

35.0

911.4

262

Ulm

us

rubra

slip

per

yel

mE

G1

3,4

91

2,4

27,5

30

964,2

13

-60.2

8824,8

00

33.9

86.4

511.8

4277

Ulm

us

sero

tin

aS

epte

mb

erel

mE

S1

16

20,3

00

773

-96.1

90

0.0

094.7

836.7

82

Ulm

us

thom

asi

iro

ckel

mE

N1

63

119,9

80

13,9

81

-88.3

510,6

44

8.8

720.1

329.3

59

Note:EG,ea

stern-gen

eral;EN,ea

stern-north;ES,ea

stern-south

;W

,western

.