September 2001 Vol 51 No 9 BioScience 765

Articles

Global change is often perceived as human-inducedmodifications in climate Indeed human activities have

undeniably altered the atmosphere and probably the climateas well (Watson et al 1998) At the same time most of theworldrsquos forests have also been extensively modified by humanuse of the land (Houghton 1994) Thus climate and land useare two prongs of human-induced global change The effectof these forces on forests is mediated by the organisms withinforests Consideration of climate land use and biologicaldiversity is key to understanding forest response to globalchange

Biological diversity refers to the variety of life at organiza-tional levels from genotypes through biomes (Franklin 1993)The responses of ecological systems to global change reflectthe organisms that are within them While ecologists havesometimes not seen the forest for the trees so to speak it isalso true that forests cannot be understood without knowl-edge of the trees and other component species It is the re-sponses of individual organisms that begin the cascade of eco-logical processes that are manifest as changes in systemproperties some of which feed back to influence climate andland use (Figure 1) Beyond its role in ecosystems biodiver-sity is invaluable to humans for foods medicines genetic in-formation recreation and spiritual renewal (Pimentel et al1997) Thus global changes that affect the distribution andabundance of organisms will affect future human well-beingand land use as well as possibly the climate

This article serves as a primer on forest biodiversity as a keycomponent of global change We first synthesize currentknowledge of interactions among climate land use and bio-diversity We then summarize the results of new analyses onthe potential effects of human-induced climate change on for-est biodiversity Our models project how possible future cli-mates may modify the distributions of environments re-quired by various species communities and biomes Currentknowledge models and funding did not allow these analy-ses to examine the population processes (eg dispersal re-generation) that would mediate the responses of organismsto environmental change It was also not possible to model theimportant effects of land use natural disturbance and otherfactors on the response of biodiversity to climate changeDespite these limitations the analyses discussed herein areamong the most comprehensive projections of climate changeeffects on forest biodiversity yet conducted We concludewith discussions of limitations research needs and strategiesfor coping with potential future global change

Andrew J Hansen (e-mail hansenmontanaedu) is an associate professor in the Ecology Department at Montana State University Bozeman MT

59717 Ronald P Neilson is a bioclimatologist with the USDA Forest Service 3200 SW Jefferson Way Corvallis Oregon 97331 Virginia H Dale

is a senior scientist in the Environmental Sciences Division at Oak Ridge National Laborator y Oak Ridge TN 37831 Curtis H Flather is a research

wildlife biologist USDA Forest Service Rocky Mountain Research Station 2150 Center Ave Bldg A Fort Collins CO 80526ndash1891 Louis Iverson is

a research landscape ecologist with the USDA Forest Service Northeastern Research Station 359 Main Road Delaware OH 43066 David J

Currie is professor and chair of the Biology Department at the University of Ottawa University of Ottawa Box 450 Station A Ottawa Ontario K1N

6N5 Sarah Shafer now a research geologist with the US Geological Survey Earth Surface Team Denver CO 80225 was a research associate in

the Department of Geography at the University of Oregon while this article was being prepared Rosamonde Cook was a postdoctoral research

associate in the Department of Fishery and Wildlife Biology at Colorado State University while this article was being prepared Patrick J Bartlein is

a professor in the Department of Geography at University of Oregon Eugene OR 97403-1251 copy 2001 American Institute of Biological Sciences

Global Change in ForestsResponses of SpeciesCommunities and BiomesANDREW J HANSEN RONALD P NEILSON VIRGINIA H DALE CURTIS H FLATHER LOUIS R IVERSONDAVID J CURRIE SARAH SHAFER ROSAMONDE COOK PATRICK J BARTLEIN

INTERACTIONS BETWEEN CLIMATE

CHANGE AND LAND USE ARE PROJECTED

TO CAUSE LARGE SHIFTS IN BIODIVERSITY

766 BioScience September 2001 Vol 51 No 9

Articles

Effects of global change on speciescommunities and biomesThis review focuses on three levels of biodiversity speciescommunities and biomes Communities are assemblages ofinteracting species such as sprucendashfir forests or tallgrassprairies Biomes are major biogeographic regions consistingof distinctive plant life forms (eg forests grasslands) Thedistributions of species reflect their differential responses toclimate edaphic factors and biotic interactions such as com-petition and herbivory These species dynamics provide themechanisms by which communities and biomes respond toglobal changeAt the same time biomes and communities mayconstrain the nature of the speciesrsquo response For examplespecies may not be able to shift range without being accom-panied by mutualists such as pollinators

The distribution and abundance of a species are governedby the birth growth death and dispersal rates of individu-als comprising a population These vital rates are in turn in-fluenced by environmental factors including climate whichalter resource availability fecundity and survivorship (Hansenand Rotella 1999) When aggregated across populationschanges in vital rates manifest as local extinction and colo-nization events which are the mechanisms by which speciesrsquoranges change Some of the best evidence that speciesrsquo dis-tributions are affected by changing climates is found amongplants The North American flora shifted during Holocenewarming and the rates and direction of response differedamong species (Webb 1992)

The birth death and other vital rates of species are also af-fected by land use Humans modify the quality amount andspatial configuration of habitats Degradation of habitatquality or quantity can reduce population size and growthrates and elevate the chance of local extinction events (Pul-liam 1988) Such habitat loss can reduce genetic diversity andthe ability of species to evolve adaptations to new environ-ments (Gilpin 1987) Land use may also alter the habitat

spatial pattern increasing the distances among habitat patchesAn important consequence of this habitat fragmentation isa reduction in habitat connectivity which could constrain theability of many species to move across the landscape in re-sponse to climate change (Primack and Miao 1992 Iversonet al 1999a)

Because of the individualistic responses of species bioticcommunities are not expected to respond to climate changeas intact units Community composition will change in re-sponse to a complex set of factors including the direct effectsof climate differential species dispersal and indirect effectsassociated with changes in disturbance regimes land useand interspecific interactions (Peters 1992) Such an indi-vidualistic perspective implies that the impacts of climatechange on communities can be understood by the aggregateresponse across species However this perspective may beflawed because interactions among species are not yet suffi-ciently understood An alternative view suggests that climatechange may affect community-level characteristics such asspecies richness and resilience to perturbation

Climate may also influence community characteristics byaltering the energy available to organismsWright et al (1993)found that species richness is often related to ecological pro-ductivity observing frequent correlations with climate foodavailability and limiting nutrients These factors have allbeen interpreted to reflect energy availability in a systemThose areas receiving more energy often have a more com-plex partitioning of usable energy among species and thus agreater richness than those areas receiving less energy (Cur-rie 1991) The influence of species richness on ecosystemfunction is complex Ecosystems with higher native speciesrichness are sometimes more resilient to perturbation (Frankand McNaughton 1991 but see also Wardle et al 2000) Alsothe presence of exotic species may elevate species richness butinhibit ecosystem function Nonetheless the diversityndashresilience relationship may be important in anticipatingecosystem response to future climates

Land-use activities also affect the availability of energy inecosystems Nearly 40 of the Earthrsquos net primary produc-tivity has been diverted to support human populations (Vi-tousek et al 1986) This change in land cover has resulted inthe conversion of native habitats supporting diverse speciesassemblages to intensive land uses that support only simpli-fied low-diversity communities (Rapport et al 1985) The im-pacts of eroding biodiversity could include reductions in re-silience resistance to invasion and ecological services providedto humans

Climate is the primary force shaping the major biomes ofthe world Mean and variation in annual precipitation andtemperature explain much of the observed pattern of biomedistribution (Prentice 1990) Shifts in biome location de-pend on the movements of key species Because the pre-dicted rates of climate change will push the climatic bound-aries of biomes northward at a rate faster than the predictedrate of species migration (Davis and Zabinski 1992) shifts inbiomes will probably lag behind changes in climate

Figure 1 Conceptual model of the aspects ofglobal change Several other factors that mayinfluence biodiversity are discussed in the textbut are not included in the modeling ofbiodiversity response to climate change

Land-use activities may appear to be too local in scale toaffect regional biome distributions However in some biomessuch as the North American Prairie land conversion is so ex-tensive that little native vegetation remains Land use canalso affect biomes to an extent greater than the sum of the areadirectly affected Conversion of natural vegetation to an-thropogenic cover types can alter the frequency and scale ofdisturbance agents that also define the character of biomesFor example fire suppression and grazing in the AmericanSouthwest are thought to be partially responsible for the in-vasion of woody vegetation into arid grassland habitats(Brown and Davis 1995) For a more compete discussion ofinteractions between climate change and disturbance seeDale et al (2001)

Feedbacks among climateland use and biodiversityClimate and land use often interact in ways that influence bio-diversity implying that these factors cannot be considered inisolation For example land use may modify climatic impactson species distributions by altering dispersal routes Whereland use creates barriers to dispersal for native species and fa-cilitates dispersal for exotic species climate change in human-dominated landscapes is likely to select for exotic species andagainst many native species (Malanson and Cairns 1997) Suchconstraints on dispersal are of concern especially around na-ture reserves Organisms ldquotrappedrdquo in reserves by surround-ing land use may become extinct if they are not able to dis-perse to increasingly suitable habitats in other nature reserves(Halpin 1997) Even in nature reserves ldquoweedyrdquo species willmost likely be quick to replace native species that succumb toclimate change

Climate and land use also jointly influence disturbancessuch as wildfire flooding and landslides Some land usespreset the landscape to be very sensitive to extreme climateevents leading to severe disturbance Logging can cause dry-ing of fuels and allow severe fires during normal drought pe-riods (Franklin and Forman 1987) This situation is thoughtto have transpired in Indonesia where slash-and-burn agri-cultural practices provided fuels and ignition sources whenan El Nintildeo event induced drought in 1997 Consequently vastareas of the rainforest burned possibly jeopardizing many en-demic species Roads and logging practices can similarly in-crease land sliding and flooding during storm events (Swan-son and Dryness 1975) Livestock grazing on the other handoften reduces fuel loads and reduces wildfire frequency andintensity for a given climate condition (Arno and Gruell1983)

Changes in land-cover patterns can also directly affect cli-mate which in turn influences biodiversity (Dale 1997) Forexample deforestation over large areas can cause reductionsin transpiration cloud formation and rainfall and increasedlevels of drying (Dickenson 1991) Such changes can lead todramatic shifts in biome type such as the replacement offorests by shrubs or grassland Similarly agricultural landuse and irrigation in the western Great Plains has been asso-

ciated with increased cloudiness and precipitation in theRocky Mountains of Colorado (Chase et al 1999)

Recent responses of biodiversity to global changeTrends in biodiversity to changes in climate and land useover the last 100 years provide a context for understanding fu-ture interactions During the past century average air tem-peratures have increased 0ndash2 ordmC over most of the UnitedStates and Canada (Watson et al 1998) Changes in precip-itation have been variable over this period increasing from5 to 20 over most of the United States but decreasing upto 20 in the Northern Rockies and California (Watson et al1998) The US population grew from 76 million in 1900 toover 270 million in 1998 Agricultural lands increased in areauntil 1900 then decreased slowly until 1950 they have re-mained stable since then (Maizel et al 1999) Between 1942and 1992 urban area increased 120 while protected areasincreased 80 (Flather et al 1999) Rural residential devel-opment has increased rapidly in the mountain west in recentdecades expanding human influences into semi-natural habi-tats

Many species and communities have responded to thesechanges in climate and land use For example forest declineand dieback are evident along the Atlantic and Pacific coastsand may be related to elevated CO2 levels and climate change(Mueller-Dombois 1992) The breeding ranges of some mo-bile species such as waterfowl have been expanding north-ward in association with climate amelioration (Abrahamand Jefferies 1997) Shifts in demography are apparent insome species Both amphibians and birds in Great Britain haveshifted breeding dates by 1 to 3 weeks earlier since the 1970sin association with increasing temperatures (Beebee 1995Crick et al 1997)

Species associated with human-dominated landscapeshave greatly expanded in recent years These include manylarge ungulate and small mammal species waterfowl andsome bird species associated with agricultural and urban en-vironments (Flather et al 1999) Some of these species are nowso abundant (eg deer) that the primary concern is controllingtheir populations Many exotic species have also become es-tablished in the United States and greatly expanded theirranges (Drake et al 1989)

At the same time several natural community types and nu-merous species have been greatly reduced by human activi-ties For example natural sprucendashfir longleaf pine andloblollyndashshortleaf pine forests now cover less than 2 oftheir presettlement ranges (Noss et al 1994) and are likely tobe further reduced under global warming Many species de-pendent upon these endangered and reduced ecosystems arecurrently in peril The number of species listed as threat-ened and endangered in the United States under the Endan-gered Species Act currently totals 1232 (USDI 2000) Factorscontributing to species endangerment include habitat con-version resource extraction and exotic species (Wilcove et al1998) The spatial distribution of such factors results in

September 2001 Vol 51 No 9 BioScience 767

Articles

species at risk being concentrated in particular regions ofthe United States especially the southern Appalachians thearid Southwest and coastal areas (Flather et al 1998)

Biodiversity under future global changeGiven past relationships among climate land use and bio-diversity how might biodiversity respond to future globalchange We assess potential vegetation response to projectedfuture climate change for doubled CO2 concentrations by us-ing the climate predictions of general circulation models(GCMs) (Table 1) as input to a set of different vegetation sim-ulation models The primary focus is on the potential conti-nental-scale response of forest vegetation as reflected inchanges in the distributions of biomes community types andtree species Although vegetation models incorporating land-use dynamics are under rapid development at local scales thecurrent state of knowledge does not allow for integrating theeffects of land use at the continental scale In our assessmentwe chose to emphasize forests as a key element of biodiver-sity which allowed us to draw implications for other organ-isms that require forest habitats We also simulate changes inspecies richness of trees and terrestrial vertebrates based onenergy theory (Currie 1991) and examine potential effects ofclimate change on locations where endangered species are con-centrated Results are for the coterminous United States un-less stated otherwise Each of these assessments is summarizedhere and reported in detail elsewhere (Table 2)

The models used to anticipate biodiversity responses tochanges in climate are based on empirical relationships be-tween organisms (vegetation and animals) and the environ-ments they currently occupy By using these relationships to

predict biodiversity response we are implicitly assuming thatthese environmentndashorganism relationships will remain un-altered in the future Although this approach can be criticizedfor not modeling some or all of the actual mechanisms lead-ing to shifts in vegetation and species ranges such an approachis commonly taken (Rogers and Randolph 2000) to initiallyassess ecological responses to global change scenarios It is alsoimportant to emphasize that the given GCMs are coarse-grid regionally smoothed outputs that do not allow depictionof local or even subregional climates Consequently the vegetation outputs modeled here must also be consideredcoarse and not sensitive to local phenomena

Climate change scenarios The GCMs differ in formu-lation hence predictions vary among the models Conse-quently the simulations are best considered as possible al-ternative views of the future with unknown likelihood ofoccurrence Both equilibrium and transient GCM scenariosare used in this assessment to incorporate a broad array of pos-sible futures (see Aber et al 2001) We put greater confidenceinto outcomes for which the models are in agreement we takedisagreement among the models as an indication of uncer-tainty Thus we report major findings for which most of themodels agree and we point out disagreement Still the un-certainty level for each climate and vegetation output is un-known and probably high given the uncertainty in fore-casting climate and subsequent vegetation responses

For the coterminous United States the different climate sce-narios all predict some level of warming and increased annualprecipitation (Table 1) Mean annual temperature increasesvary from 33degC to 58degC with the greatest warming at higherlatitudes Mean annual precipitation is predicted to increase

768 BioScience September 2001 Vol 51 No 9

Articles

Table 1 General circulation models and predictions for change from current average annual temperature (dT) andprecipitation (dP) for the coterminous United States under a doubling of atmospheric CO2 The equilibrium-type modelssimulate an instantaneous increase in CO2 and are run until equilibrium climate conditions emerge The newer transientmodels assume trace gases increase at 1 per year until 2100 and allow the climate to adjust while incorporatinginherent lags in the oceanndashatmosphere systems The HADCM2SUL and CGCM1 scenarios include the effects of sulfateaerosols The dT and dP values from the transient scenarios were calculated from averages of the last 30 years of thescenarios compared with the 1961ndash1990 means

Name Acronym Type Reference dT (oC) dP ()

Oregon State University OSU Equilibrium Schlesinger and Zhao 1989 3 21

Geophysical Fluids Dynamics Laborator y GFDLR30 Equilibrium Manabe et al 1990 42 189

Goddard Institute for Space Studies GISS Equilibrium Hansen et al 1988 44 51

United Kingdom Meteorological Office UKMO Equilibrium Wilson and Mitchell 1987 66 113

UKMO Hadley Centre HADCM2SUL Transient Johns et al 1997 28 229

UKMO Hadley Centre HADCM2GHG Transient Johns et al 1997 37 307

Canadian Climate Centre CGCM1 Transient Boer et al 2000 52 215

in the West with 20 to more than 50 increases in Cali-fornia Decreased precipitation of up to 30 is predicted forlocations in the Southeast Texas and the Northwest

Biomes The MAPSS biogeography model (Neilson 1995)projects biome response to climate as change in vegetationstructure and density based on light water and nutrient lim-itations (VEMAP members 1995 Bachelet et al 2001) Veg-etation is coupled directly to climate and hydrology andrules are applied to classify vegetation into biome types Themodel considers the effects of altered CO2 on plant physiol-ogy MAPSS simulates potential natural vegetation (specifi-cally life-forms) based on climate and does not include suc-cession or plant dispersal

The results project that potential forest area decreases byan average of 11 across the GCM scenarios with a range of+23 under the coolest scenarios and ndash45 under the hottestscenarios (Figure 2) Northeast mixed (hardwood and conifer)forests decrease by 72 in potential area on average (rangendash14 to ndash97) as they shift into Canada and increase in po-tential area continentally (Watson et al 1998) The potentialarea of eastern hardwoods decreases by an average of 34(range ndash93 to +51) These deciduous forests shift northreplacing northeastern mixed forests but are squeezed fromthe south by southeastern mixed forests or from the west bysavannas and grasslands depending on the scenario The po-tential range of southeastern mixed forests increases under allscenarios (average 37 range 25 to 57) while shiftingnorth This biome remains intact under cooler scenarios butis converted to savannas and grasslands in the South underthe hotter scenarios

Potential environments for alpine ecosystems all but dis-appear from the western mountains being overtaken by en-croaching forests Wet coniferous forests in the Northwest de-crease in potential area by 9 on average (range ndash54 to+21) The potential range of interior western pines change

little on average with a range of ndash30 to +39 The poten-tial range of shrublands and arid woodlands expand in the in-terior West and Great Plains encroaching on some grasslandsHowever grassland habitats may expand in the deserts ofthe Southwest parts of the Southeast and possibly in the up-per Midwest Thus the potential area of grassland could ei-ther decrease or increase

The projections agree on a single or on two vegetationclasses across 68 of the coterminous United States Locationsof greatest certainty are in the northern plains and FloridaRegions of great uncertainty are transition zones in the east-ern prairie and the West Taken in order of increasing tem-perature change the future scenarios imply that potentialforest range could expand with small amounts of warming butwould contract under the hotter scenarios

Forest community types and tree species Statisticalmodels are used to project the distributions of tree species withresults aggregated into community types These models pro-ject tree response to future climate based on current rela-tionships between trees and environmental variables such asclimate and soils Because physiological data are not required(as is the case for process models) many species can be mod-eled However these approaches do not include important for-est dynamics involving species interactions physiologicalprocesses such as CO2 uptake and tree dispersal and estab-lishment They also assume that speciesndashenvironment rela-tionships will remain the same under future climate condi-tions which may not be the case

Two statistical models are used The DISTRIB model ofIverson and Prasad (Iverson and Prasad 1998 Iverson et al1999a Prasad and Iverson 1999) was applied to the easternUnited States This model uses regression tree analysisbased on 33 environmental variables to predict the poten-tial future distribution of suitable habitat for 80 tree speciesAn index of species response (regional importance) was

September 2001 Vol 51 No 9 BioScience 769

Articles

Table 2 Biodiversity models used in this analysis

Response variables Model Model type Resolution Extent GCMs simulated Reference

Biomes community MAPSS Biogeographic 10 km grid coterminous HADCM2S HADCM2G Neilson 1995types processes United States CGCM1 OSU Bachelet et al

GFDLndashR30 GISS 2001UKMO

Tree species forest DISTRIB Statistical County Eastern United HADCM2S CGCM1 Iverson and community types regression tree States GFDLndashR30 GISS UKMO Prasad 1998

2001

Tree and shrub Response Statistical 25 km grid North America HADCM2S CGCM1 Shafer et al 2001species surface local regression

model

Species richness of Currie model Statistical 25deg x 25deg lat United States CGCM1 OSU GFDLndashR30 Currie 2001trees mammals birds and long south and Canada GISS UKMOreptiles amphibians of 50degN 5deg

long x 25deg lat north of 50degN

derived by multiplying the importance value (reflecting therelative abundance of a species in a community) by thearea for each county Community types were then definedby simply aggregating the importance values for individualtree species (Iverson and Prasad 2001) For western forestswe drew on the work of Shafer et al (2001) and Bartlein etal (1997) who use a local regression model to predict prob-ability of occurrence of tree species across North Americabased on climate and soils The results for dominant west-ern species are presented

In the eastern United States the area of habitat suitable foroakndashhickory expands in area following climate change by anaverage of 34 primarily to the north and east (Figure 3) Theoakndashpine habitat also expands by roughly 290 and is rep-resented throughout the Southeast On the other hand the

habitat capable of supporting the sprucendashfir and aspenndashbirchtypes are dramatically reduced (ndash97 and ndash92) and arelargely replaced by oakndashhickory and oakndashpine habitat Theloblollyndashshortleaf pine habitat is also reduced by 32 itshifts to the north and west while being replaced in its cur-rent zone by oakndashpine habitat The longleafndashslash pine habi-tat is reduced on average by 31 There is a higher level ofdisagreement among climate scenarios for the elmndashashndashcot-tonwood oakndashgumndashcypress and whitendashredndashjack pine typeswhich show increases in habitat under some scenarios and de-creases under others

These potential changes in habitat for community types re-flect the responses of individual tree species Seven of the 80species modeled were predicted to have their suitable habi-tat reduced in regional importance by at least 90 bigtooth

770 BioScience September 2001 Vol 51 No 9

Articles

Figure 2 Simulated potential distributions of biomes using the MAPSS biogeography model under six different GCMscenarios representing approximately 2 acute CO

2concentration (a) Color key for biomes (b) simulated current biome

distribution based on average (1961ndash1990) climate (c) uncertainty map (the number of unique biome types simulatedacross all six GCM scenarios are plotted) (d) biome distribution under the HADCM2SUL scenario among the coolest offuture warming scenarios (e) a modal map of future biome distributions (shown are the biomes most often simulated forthe future across all six GCMs refer to panel c for the ldquouncertaintyrdquo associated with the modal map) (f) biome distributionunder the CGCM1 scenario among the warmest of future scenarios

aspen (Populus grandiden-tata) quaking aspen (Ptremuloides ) sugar maple(Acer saccharum) northernwhite cedar (Thuja occiden-talis) balsam fir (Abies bal-samea) red pine (Pinusresinosa) and paper birch(Betula papyrifera) An ad-ditional 24 species would po-tentially decline by at least10 In contrast regionalimportance increased for 35species including 12 speciesthat increased suitable habi-tat by 100 or more in-cluding four species of oakand one hickory

Most speciesrsquo habitat wasprojected to move to thenorth 100 to 530 km for sev-eral species Some speciessuch as quaking aspen paperbirch northern white cedarbalsam fir and sugar maplehave the optimum latitudeof suitable habitat movenorth of the US border

All of the above analysesfor the eastern United Statesrelate to the potential distri-bution of suitable habitatnot actual distribution of thespecies The assumption isthat the species will get therethat there are no barriers orconstraints to migration(For newer projections thatconsider dispersal see Iver-son et al 1999b)

In the western UnitedStates the potential rangesof dominant rainforestconifers such as westernhemlock are projected to de-crease west of the CascadeMountains and expand intomountain ranges through-out the interior West Simu-lated potential habitat forDouglas fir (Pseudotsugamenziesii) also decreasesalong the western coast ofthe coterminous UnitedStates but expands east ofthe Cascades and Sierras as

September 2001 Vol 51 No 9 BioScience 771

Articles

Figure 3 Simulated potential distributions of forest community types in the eastern UnitedStates using the DISTRIB model under five different GMC scenarios representingapproximately 2 x CO

2concentration (a) Color key for forest types (b) current forest type

distribution based on 10000 actual forest inventory plots (c) uncertainty map (plotted arethe number of unique forest community types simulated across all five future GCM scenarios)(d) potential forest community type distribution under the HADCM2SUL scenario amongthe coolest of future warming scenarios (e) a modal map of future biome distributions(shown are the biomes most often simulated for the future across all five GMCs refer to panel[c] for the ldquouncertaintyrdquo associated with the modal map) and (f) forest community typedistribution under the CGCM1 scenario among the warmest of future scenarios

well as northward along thewest coast of Canada intoAlaska (Figure 4)

The potential range of sev-eral subalpine conifers is sim-ulated to contract substantiallyin the western coterminousUnited States including En-glemann spruce (Picea engel-mannii) mountain hemlock(Tsuga mertensiana) and sev-eral species of true fir (Abies)Potential ranges for these sub-alpine species are simulated toexpand along the western coastof Canada and into AlaskaPotential future habitat for bigsagebrush (Artemisia triden-tata) an important shrub inthe inland West is largely ab-sent in the United States shift-ing into Canada This shrubis simulated to be replaced inthe United States by shrubschaparral and grasslands nowfound in arid regions of theSouthwest Potential habitatfor ponderosa pine (Pinusponderosa) is simulated to ex-pand in the western UnitedStates including on the WestCoast where many otherconifers are projected to con-tract

While the potential rangesof many taxa in the West shiftnorthward the topographiccomplexity of the region cre-ates less intuitive changes insome species distributions Po-tential habitat for some coniferspecies associated with mesicclimates shifts south and eastalong the Rocky Mountainswith for example forests typ-ical today of Glacier NationalPark becoming dominant inYellowstone National Park tothe southeast (Bartlein et al1997) The direction of changemay also differ from east towest within a speciesrsquo rangePotential habitat for paperbirch whose range spans thecontinent is simulated to con-tract northward in the eastern

772 BioScience September 2001 Vol 51 No 9

Articles

Figure 4 Simulated distributions and scenario agreement for Betula papyriferaPseudotsuga menziesii Pinus ponderosa and Artemisia tridentata Estimatedprobabilities of occurrence for a taxon simulated with observed modern climate (leftpanel) Comparison of the observed distributions with the simulated distributions underfuture climate conditions as generated by HADCM2S and CGCM1 for 2090ndash2099 (middlepanels) Gray indicates locations where the taxon is observed today and is simulated tooccur under future climate conditions red indicates locations where the taxon is observedtoday but is simulated to be absent under future climate conditions and blue indicateslocations where the taxon is absent today but is simulated to occur under future climateconditions Scenario agreement (right panel) Light purple indicates locations where thespecies is simulated to be present under the future climate of either the HADCM2S orCGCM1 scenario dark purple indicates locations where the species is simulated to bepresent under both future climate scenarios

United States but to expand southward along the RockyMountains in the West (Figure 4) The complex topographyof the West results in simulated future habitat for many treespecies that is disjunct Consequently dispersal to new habi-tats under climate change may be more difficult in the Westthan in the East where the future distributions of species aresimulated to be more continuous

Community richness Currie (2001) derived multipleregression models relating the broad-scale variability ofspecies richness of trees and terrestrial vertebrates acrossNorth America to the spatial patterns of summer and winterprecipitation and temperature These models were used to pre-dict patterns of species richness under five GCM scenariosThis approach assumes that richness will continue to covarywith climate in the future in the same way that it does today

The results indicate that contemporary patterns of richnesscorrelate strongly with temperature and less strongly with pre-cipitation These climate-richness relationships differ amongtaxonomic groups thus projected change in richness underclimate change also differed among groups Current treespecies richness is a positive function of temperature up to rel-atively high temperatures then is negatively related to furtherincreases in temperature Current tree species richness is alsoa positive function of precipitation Thus climatic warmingis predicted to lead to increased tree richness over most of thenorthern United States especially in the areas that are nowcoldest in the western mountains and near the CanadianBorder (Figure 5) Moderate decreases in richness (ndash20) arepredicted to occur in areas that are likely to experience dry-ing and very high temperatures such as the southwesterndeserts The greatest disagreement among the projectionswas for Pacific Northwest All models predict increases inrichness in that area but these increases may be modest (lt +50) to pronounced (gt +100)

Contemporary species richness of endotherms (birds andmammals) covaries strongly with temperature Richness inthese groups is maximal in moderately warm areas (thesouthern Appalachians and southern Rockies) and it de-creases in hotter areas This relationship may occur becauseambient heat serves as a direct energy subsidy for endothermsin cold climates but these organisms expend energy to dis-sipate heat in hot areas Endotherm richness is only weaklyrelated to precipitation Consequently under most climatechange scenarios endotherm richness is predicted to de-crease by over 25 in low-elevation areas in the SoutheastIncreases in richness (gt +11 to gt +100) are predicted forupper montane areas across the United States

Contemporary ectotherm (reptiles and amphibians) rich-ness is even more strongly related to temperature increasingmonotonically as temperature increases Ambient heat is alsoan energy subsidy for ectotherms even in the warmest areasof the earth Consequently climatic warming is predicted toincrease ectotherm richness over the entire coterminousUnited States For reptiles the increase is predicted to bemodest across the South and greater in the North Amphib-

ian richness in contrast also depends somewhat on precip-itation Because most GCMs predict that the southeasternUnited States will become somewhat drier in winter am-phibian richness in the Southeast is predicted to change lit-tle despite increased temperatures and to increase elsewhereDifferences among predictions of different GCMs lie mainlyin how dramatically richness is likely to increase

Threatened and endangered species How might thechanges in species richness predicted above influence en-dangered species hotspots (places where many endangeredspecies occur) We overlaid the maps of projected species rich-ness over maps of current hotspots for threatened and en-dangered species (Flather et al 1998) to determine the pro-portion of each endangerment hotspot area in the contiguousUnited States that is predicted to show an increase decreaseor no change in species richness for each taxonomic groupThe results indicated that reptiles and amphibians are expectedto increase in richness across all endangerment hotspots Onthe other hand bird and mammal richness may undergosignificant reductions in many endangerment hotspots es-pecially in the East (Figure 6)

Implications of assessment resultsIn contrast to public perceptions of large-scale forest lossunder global change the biome models project only a mod-est average loss of forest area (11) for the coterminousUnited States Lost forest is largely replaced by savanna andarid woodland biome types However the projected responseof forest habitats to global change scenarios is highly variableamong GCMs Forest habitats are projected to increase by 23under the HADCM2SUL model and are predicted to de-crease by 45 under the UKMO model This uncertaintypoints to cautious interpretation of our findings and to theneed for further research directed at understanding the fac-tors driving change in forest systems

Within the climate projections considered we did observesome common patterns in the response of forest communitytypes across the continent Areal expansion of habitats was pro-jected for oakndashhickory and oakndashpine in the East and ponderosapine and arid-tolerant hardwoods in the West The first threeof these are extremely valuable for forest products and ashabitat The heavy mast production of oakndashhickory for ex-ample is a food source for more than 180 different kinds ofvertebrates (Rogers 1990) The oak types also produce per-sistent coarse woody debris that benefits several ecologicalprocesses and organisms

Suitable habitats for several important community typeshowever are projected to greatly decrease in area or disappearfrom the coterminous United States These include alpine habi-tats subalpine sprucendashfir forests aspen the maplendashbeechndashbirchtype sagebrush and loblollyndashshortleaf pine communitiesSubalpine sprucendashfir has been decreasing in area becomingincreasingly fragmented and losing species richness sincethe last glacial period (Brown and Davis 1995) Other habi-tats such as sagebrush and aspen are being reduced in mod-

September 2001 Vol 51 No 9 BioScience 773

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

766 BioScience September 2001 Vol 51 No 9

Articles

Effects of global change on speciescommunities and biomesThis review focuses on three levels of biodiversity speciescommunities and biomes Communities are assemblages ofinteracting species such as sprucendashfir forests or tallgrassprairies Biomes are major biogeographic regions consistingof distinctive plant life forms (eg forests grasslands) Thedistributions of species reflect their differential responses toclimate edaphic factors and biotic interactions such as com-petition and herbivory These species dynamics provide themechanisms by which communities and biomes respond toglobal changeAt the same time biomes and communities mayconstrain the nature of the speciesrsquo response For examplespecies may not be able to shift range without being accom-panied by mutualists such as pollinators

The distribution and abundance of a species are governedby the birth growth death and dispersal rates of individu-als comprising a population These vital rates are in turn in-fluenced by environmental factors including climate whichalter resource availability fecundity and survivorship (Hansenand Rotella 1999) When aggregated across populationschanges in vital rates manifest as local extinction and colo-nization events which are the mechanisms by which speciesrsquoranges change Some of the best evidence that speciesrsquo dis-tributions are affected by changing climates is found amongplants The North American flora shifted during Holocenewarming and the rates and direction of response differedamong species (Webb 1992)

The birth death and other vital rates of species are also af-fected by land use Humans modify the quality amount andspatial configuration of habitats Degradation of habitatquality or quantity can reduce population size and growthrates and elevate the chance of local extinction events (Pul-liam 1988) Such habitat loss can reduce genetic diversity andthe ability of species to evolve adaptations to new environ-ments (Gilpin 1987) Land use may also alter the habitat

spatial pattern increasing the distances among habitat patchesAn important consequence of this habitat fragmentation isa reduction in habitat connectivity which could constrain theability of many species to move across the landscape in re-sponse to climate change (Primack and Miao 1992 Iversonet al 1999a)

Because of the individualistic responses of species bioticcommunities are not expected to respond to climate changeas intact units Community composition will change in re-sponse to a complex set of factors including the direct effectsof climate differential species dispersal and indirect effectsassociated with changes in disturbance regimes land useand interspecific interactions (Peters 1992) Such an indi-vidualistic perspective implies that the impacts of climatechange on communities can be understood by the aggregateresponse across species However this perspective may beflawed because interactions among species are not yet suffi-ciently understood An alternative view suggests that climatechange may affect community-level characteristics such asspecies richness and resilience to perturbation

Climate may also influence community characteristics byaltering the energy available to organismsWright et al (1993)found that species richness is often related to ecological pro-ductivity observing frequent correlations with climate foodavailability and limiting nutrients These factors have allbeen interpreted to reflect energy availability in a systemThose areas receiving more energy often have a more com-plex partitioning of usable energy among species and thus agreater richness than those areas receiving less energy (Cur-rie 1991) The influence of species richness on ecosystemfunction is complex Ecosystems with higher native speciesrichness are sometimes more resilient to perturbation (Frankand McNaughton 1991 but see also Wardle et al 2000) Alsothe presence of exotic species may elevate species richness butinhibit ecosystem function Nonetheless the diversityndashresilience relationship may be important in anticipatingecosystem response to future climates

Land-use activities also affect the availability of energy inecosystems Nearly 40 of the Earthrsquos net primary produc-tivity has been diverted to support human populations (Vi-tousek et al 1986) This change in land cover has resulted inthe conversion of native habitats supporting diverse speciesassemblages to intensive land uses that support only simpli-fied low-diversity communities (Rapport et al 1985) The im-pacts of eroding biodiversity could include reductions in re-silience resistance to invasion and ecological services providedto humans

Climate is the primary force shaping the major biomes ofthe world Mean and variation in annual precipitation andtemperature explain much of the observed pattern of biomedistribution (Prentice 1990) Shifts in biome location de-pend on the movements of key species Because the pre-dicted rates of climate change will push the climatic bound-aries of biomes northward at a rate faster than the predictedrate of species migration (Davis and Zabinski 1992) shifts inbiomes will probably lag behind changes in climate

Figure 1 Conceptual model of the aspects ofglobal change Several other factors that mayinfluence biodiversity are discussed in the textbut are not included in the modeling ofbiodiversity response to climate change

Land-use activities may appear to be too local in scale toaffect regional biome distributions However in some biomessuch as the North American Prairie land conversion is so ex-tensive that little native vegetation remains Land use canalso affect biomes to an extent greater than the sum of the areadirectly affected Conversion of natural vegetation to an-thropogenic cover types can alter the frequency and scale ofdisturbance agents that also define the character of biomesFor example fire suppression and grazing in the AmericanSouthwest are thought to be partially responsible for the in-vasion of woody vegetation into arid grassland habitats(Brown and Davis 1995) For a more compete discussion ofinteractions between climate change and disturbance seeDale et al (2001)

Feedbacks among climateland use and biodiversityClimate and land use often interact in ways that influence bio-diversity implying that these factors cannot be considered inisolation For example land use may modify climatic impactson species distributions by altering dispersal routes Whereland use creates barriers to dispersal for native species and fa-cilitates dispersal for exotic species climate change in human-dominated landscapes is likely to select for exotic species andagainst many native species (Malanson and Cairns 1997) Suchconstraints on dispersal are of concern especially around na-ture reserves Organisms ldquotrappedrdquo in reserves by surround-ing land use may become extinct if they are not able to dis-perse to increasingly suitable habitats in other nature reserves(Halpin 1997) Even in nature reserves ldquoweedyrdquo species willmost likely be quick to replace native species that succumb toclimate change

Climate and land use also jointly influence disturbancessuch as wildfire flooding and landslides Some land usespreset the landscape to be very sensitive to extreme climateevents leading to severe disturbance Logging can cause dry-ing of fuels and allow severe fires during normal drought pe-riods (Franklin and Forman 1987) This situation is thoughtto have transpired in Indonesia where slash-and-burn agri-cultural practices provided fuels and ignition sources whenan El Nintildeo event induced drought in 1997 Consequently vastareas of the rainforest burned possibly jeopardizing many en-demic species Roads and logging practices can similarly in-crease land sliding and flooding during storm events (Swan-son and Dryness 1975) Livestock grazing on the other handoften reduces fuel loads and reduces wildfire frequency andintensity for a given climate condition (Arno and Gruell1983)

Changes in land-cover patterns can also directly affect cli-mate which in turn influences biodiversity (Dale 1997) Forexample deforestation over large areas can cause reductionsin transpiration cloud formation and rainfall and increasedlevels of drying (Dickenson 1991) Such changes can lead todramatic shifts in biome type such as the replacement offorests by shrubs or grassland Similarly agricultural landuse and irrigation in the western Great Plains has been asso-

ciated with increased cloudiness and precipitation in theRocky Mountains of Colorado (Chase et al 1999)

Recent responses of biodiversity to global changeTrends in biodiversity to changes in climate and land useover the last 100 years provide a context for understanding fu-ture interactions During the past century average air tem-peratures have increased 0ndash2 ordmC over most of the UnitedStates and Canada (Watson et al 1998) Changes in precip-itation have been variable over this period increasing from5 to 20 over most of the United States but decreasing upto 20 in the Northern Rockies and California (Watson et al1998) The US population grew from 76 million in 1900 toover 270 million in 1998 Agricultural lands increased in areauntil 1900 then decreased slowly until 1950 they have re-mained stable since then (Maizel et al 1999) Between 1942and 1992 urban area increased 120 while protected areasincreased 80 (Flather et al 1999) Rural residential devel-opment has increased rapidly in the mountain west in recentdecades expanding human influences into semi-natural habi-tats

Many species and communities have responded to thesechanges in climate and land use For example forest declineand dieback are evident along the Atlantic and Pacific coastsand may be related to elevated CO2 levels and climate change(Mueller-Dombois 1992) The breeding ranges of some mo-bile species such as waterfowl have been expanding north-ward in association with climate amelioration (Abrahamand Jefferies 1997) Shifts in demography are apparent insome species Both amphibians and birds in Great Britain haveshifted breeding dates by 1 to 3 weeks earlier since the 1970sin association with increasing temperatures (Beebee 1995Crick et al 1997)

Species associated with human-dominated landscapeshave greatly expanded in recent years These include manylarge ungulate and small mammal species waterfowl andsome bird species associated with agricultural and urban en-vironments (Flather et al 1999) Some of these species are nowso abundant (eg deer) that the primary concern is controllingtheir populations Many exotic species have also become es-tablished in the United States and greatly expanded theirranges (Drake et al 1989)

At the same time several natural community types and nu-merous species have been greatly reduced by human activi-ties For example natural sprucendashfir longleaf pine andloblollyndashshortleaf pine forests now cover less than 2 oftheir presettlement ranges (Noss et al 1994) and are likely tobe further reduced under global warming Many species de-pendent upon these endangered and reduced ecosystems arecurrently in peril The number of species listed as threat-ened and endangered in the United States under the Endan-gered Species Act currently totals 1232 (USDI 2000) Factorscontributing to species endangerment include habitat con-version resource extraction and exotic species (Wilcove et al1998) The spatial distribution of such factors results in

September 2001 Vol 51 No 9 BioScience 767

Articles

species at risk being concentrated in particular regions ofthe United States especially the southern Appalachians thearid Southwest and coastal areas (Flather et al 1998)

Biodiversity under future global changeGiven past relationships among climate land use and bio-diversity how might biodiversity respond to future globalchange We assess potential vegetation response to projectedfuture climate change for doubled CO2 concentrations by us-ing the climate predictions of general circulation models(GCMs) (Table 1) as input to a set of different vegetation sim-ulation models The primary focus is on the potential conti-nental-scale response of forest vegetation as reflected inchanges in the distributions of biomes community types andtree species Although vegetation models incorporating land-use dynamics are under rapid development at local scales thecurrent state of knowledge does not allow for integrating theeffects of land use at the continental scale In our assessmentwe chose to emphasize forests as a key element of biodiver-sity which allowed us to draw implications for other organ-isms that require forest habitats We also simulate changes inspecies richness of trees and terrestrial vertebrates based onenergy theory (Currie 1991) and examine potential effects ofclimate change on locations where endangered species are con-centrated Results are for the coterminous United States un-less stated otherwise Each of these assessments is summarizedhere and reported in detail elsewhere (Table 2)

The models used to anticipate biodiversity responses tochanges in climate are based on empirical relationships be-tween organisms (vegetation and animals) and the environ-ments they currently occupy By using these relationships to

predict biodiversity response we are implicitly assuming thatthese environmentndashorganism relationships will remain un-altered in the future Although this approach can be criticizedfor not modeling some or all of the actual mechanisms lead-ing to shifts in vegetation and species ranges such an approachis commonly taken (Rogers and Randolph 2000) to initiallyassess ecological responses to global change scenarios It is alsoimportant to emphasize that the given GCMs are coarse-grid regionally smoothed outputs that do not allow depictionof local or even subregional climates Consequently the vegetation outputs modeled here must also be consideredcoarse and not sensitive to local phenomena

Climate change scenarios The GCMs differ in formu-lation hence predictions vary among the models Conse-quently the simulations are best considered as possible al-ternative views of the future with unknown likelihood ofoccurrence Both equilibrium and transient GCM scenariosare used in this assessment to incorporate a broad array of pos-sible futures (see Aber et al 2001) We put greater confidenceinto outcomes for which the models are in agreement we takedisagreement among the models as an indication of uncer-tainty Thus we report major findings for which most of themodels agree and we point out disagreement Still the un-certainty level for each climate and vegetation output is un-known and probably high given the uncertainty in fore-casting climate and subsequent vegetation responses

For the coterminous United States the different climate sce-narios all predict some level of warming and increased annualprecipitation (Table 1) Mean annual temperature increasesvary from 33degC to 58degC with the greatest warming at higherlatitudes Mean annual precipitation is predicted to increase

768 BioScience September 2001 Vol 51 No 9

Articles

Table 1 General circulation models and predictions for change from current average annual temperature (dT) andprecipitation (dP) for the coterminous United States under a doubling of atmospheric CO2 The equilibrium-type modelssimulate an instantaneous increase in CO2 and are run until equilibrium climate conditions emerge The newer transientmodels assume trace gases increase at 1 per year until 2100 and allow the climate to adjust while incorporatinginherent lags in the oceanndashatmosphere systems The HADCM2SUL and CGCM1 scenarios include the effects of sulfateaerosols The dT and dP values from the transient scenarios were calculated from averages of the last 30 years of thescenarios compared with the 1961ndash1990 means

Name Acronym Type Reference dT (oC) dP ()

Oregon State University OSU Equilibrium Schlesinger and Zhao 1989 3 21

Geophysical Fluids Dynamics Laborator y GFDLR30 Equilibrium Manabe et al 1990 42 189

Goddard Institute for Space Studies GISS Equilibrium Hansen et al 1988 44 51

United Kingdom Meteorological Office UKMO Equilibrium Wilson and Mitchell 1987 66 113

UKMO Hadley Centre HADCM2SUL Transient Johns et al 1997 28 229

UKMO Hadley Centre HADCM2GHG Transient Johns et al 1997 37 307

Canadian Climate Centre CGCM1 Transient Boer et al 2000 52 215

in the West with 20 to more than 50 increases in Cali-fornia Decreased precipitation of up to 30 is predicted forlocations in the Southeast Texas and the Northwest

Biomes The MAPSS biogeography model (Neilson 1995)projects biome response to climate as change in vegetationstructure and density based on light water and nutrient lim-itations (VEMAP members 1995 Bachelet et al 2001) Veg-etation is coupled directly to climate and hydrology andrules are applied to classify vegetation into biome types Themodel considers the effects of altered CO2 on plant physiol-ogy MAPSS simulates potential natural vegetation (specifi-cally life-forms) based on climate and does not include suc-cession or plant dispersal

The results project that potential forest area decreases byan average of 11 across the GCM scenarios with a range of+23 under the coolest scenarios and ndash45 under the hottestscenarios (Figure 2) Northeast mixed (hardwood and conifer)forests decrease by 72 in potential area on average (rangendash14 to ndash97) as they shift into Canada and increase in po-tential area continentally (Watson et al 1998) The potentialarea of eastern hardwoods decreases by an average of 34(range ndash93 to +51) These deciduous forests shift northreplacing northeastern mixed forests but are squeezed fromthe south by southeastern mixed forests or from the west bysavannas and grasslands depending on the scenario The po-tential range of southeastern mixed forests increases under allscenarios (average 37 range 25 to 57) while shiftingnorth This biome remains intact under cooler scenarios butis converted to savannas and grasslands in the South underthe hotter scenarios

Potential environments for alpine ecosystems all but dis-appear from the western mountains being overtaken by en-croaching forests Wet coniferous forests in the Northwest de-crease in potential area by 9 on average (range ndash54 to+21) The potential range of interior western pines change

little on average with a range of ndash30 to +39 The poten-tial range of shrublands and arid woodlands expand in the in-terior West and Great Plains encroaching on some grasslandsHowever grassland habitats may expand in the deserts ofthe Southwest parts of the Southeast and possibly in the up-per Midwest Thus the potential area of grassland could ei-ther decrease or increase

The projections agree on a single or on two vegetationclasses across 68 of the coterminous United States Locationsof greatest certainty are in the northern plains and FloridaRegions of great uncertainty are transition zones in the east-ern prairie and the West Taken in order of increasing tem-perature change the future scenarios imply that potentialforest range could expand with small amounts of warming butwould contract under the hotter scenarios

Forest community types and tree species Statisticalmodels are used to project the distributions of tree species withresults aggregated into community types These models pro-ject tree response to future climate based on current rela-tionships between trees and environmental variables such asclimate and soils Because physiological data are not required(as is the case for process models) many species can be mod-eled However these approaches do not include important for-est dynamics involving species interactions physiologicalprocesses such as CO2 uptake and tree dispersal and estab-lishment They also assume that speciesndashenvironment rela-tionships will remain the same under future climate condi-tions which may not be the case

Two statistical models are used The DISTRIB model ofIverson and Prasad (Iverson and Prasad 1998 Iverson et al1999a Prasad and Iverson 1999) was applied to the easternUnited States This model uses regression tree analysisbased on 33 environmental variables to predict the poten-tial future distribution of suitable habitat for 80 tree speciesAn index of species response (regional importance) was

September 2001 Vol 51 No 9 BioScience 769

Articles

Table 2 Biodiversity models used in this analysis

Response variables Model Model type Resolution Extent GCMs simulated Reference

Biomes community MAPSS Biogeographic 10 km grid coterminous HADCM2S HADCM2G Neilson 1995types processes United States CGCM1 OSU Bachelet et al

GFDLndashR30 GISS 2001UKMO

Tree species forest DISTRIB Statistical County Eastern United HADCM2S CGCM1 Iverson and community types regression tree States GFDLndashR30 GISS UKMO Prasad 1998

2001

Tree and shrub Response Statistical 25 km grid North America HADCM2S CGCM1 Shafer et al 2001species surface local regression

model

Species richness of Currie model Statistical 25deg x 25deg lat United States CGCM1 OSU GFDLndashR30 Currie 2001trees mammals birds and long south and Canada GISS UKMOreptiles amphibians of 50degN 5deg

long x 25deg lat north of 50degN

derived by multiplying the importance value (reflecting therelative abundance of a species in a community) by thearea for each county Community types were then definedby simply aggregating the importance values for individualtree species (Iverson and Prasad 2001) For western forestswe drew on the work of Shafer et al (2001) and Bartlein etal (1997) who use a local regression model to predict prob-ability of occurrence of tree species across North Americabased on climate and soils The results for dominant west-ern species are presented

In the eastern United States the area of habitat suitable foroakndashhickory expands in area following climate change by anaverage of 34 primarily to the north and east (Figure 3) Theoakndashpine habitat also expands by roughly 290 and is rep-resented throughout the Southeast On the other hand the

habitat capable of supporting the sprucendashfir and aspenndashbirchtypes are dramatically reduced (ndash97 and ndash92) and arelargely replaced by oakndashhickory and oakndashpine habitat Theloblollyndashshortleaf pine habitat is also reduced by 32 itshifts to the north and west while being replaced in its cur-rent zone by oakndashpine habitat The longleafndashslash pine habi-tat is reduced on average by 31 There is a higher level ofdisagreement among climate scenarios for the elmndashashndashcot-tonwood oakndashgumndashcypress and whitendashredndashjack pine typeswhich show increases in habitat under some scenarios and de-creases under others

These potential changes in habitat for community types re-flect the responses of individual tree species Seven of the 80species modeled were predicted to have their suitable habi-tat reduced in regional importance by at least 90 bigtooth

770 BioScience September 2001 Vol 51 No 9

Articles

Figure 2 Simulated potential distributions of biomes using the MAPSS biogeography model under six different GCMscenarios representing approximately 2 acute CO

2concentration (a) Color key for biomes (b) simulated current biome

distribution based on average (1961ndash1990) climate (c) uncertainty map (the number of unique biome types simulatedacross all six GCM scenarios are plotted) (d) biome distribution under the HADCM2SUL scenario among the coolest offuture warming scenarios (e) a modal map of future biome distributions (shown are the biomes most often simulated forthe future across all six GCMs refer to panel c for the ldquouncertaintyrdquo associated with the modal map) (f) biome distributionunder the CGCM1 scenario among the warmest of future scenarios

aspen (Populus grandiden-tata) quaking aspen (Ptremuloides ) sugar maple(Acer saccharum) northernwhite cedar (Thuja occiden-talis) balsam fir (Abies bal-samea) red pine (Pinusresinosa) and paper birch(Betula papyrifera) An ad-ditional 24 species would po-tentially decline by at least10 In contrast regionalimportance increased for 35species including 12 speciesthat increased suitable habi-tat by 100 or more in-cluding four species of oakand one hickory

Most speciesrsquo habitat wasprojected to move to thenorth 100 to 530 km for sev-eral species Some speciessuch as quaking aspen paperbirch northern white cedarbalsam fir and sugar maplehave the optimum latitudeof suitable habitat movenorth of the US border

All of the above analysesfor the eastern United Statesrelate to the potential distri-bution of suitable habitatnot actual distribution of thespecies The assumption isthat the species will get therethat there are no barriers orconstraints to migration(For newer projections thatconsider dispersal see Iver-son et al 1999b)

In the western UnitedStates the potential rangesof dominant rainforestconifers such as westernhemlock are projected to de-crease west of the CascadeMountains and expand intomountain ranges through-out the interior West Simu-lated potential habitat forDouglas fir (Pseudotsugamenziesii) also decreasesalong the western coast ofthe coterminous UnitedStates but expands east ofthe Cascades and Sierras as

September 2001 Vol 51 No 9 BioScience 771

Articles

Figure 3 Simulated potential distributions of forest community types in the eastern UnitedStates using the DISTRIB model under five different GMC scenarios representingapproximately 2 x CO

2concentration (a) Color key for forest types (b) current forest type

distribution based on 10000 actual forest inventory plots (c) uncertainty map (plotted arethe number of unique forest community types simulated across all five future GCM scenarios)(d) potential forest community type distribution under the HADCM2SUL scenario amongthe coolest of future warming scenarios (e) a modal map of future biome distributions(shown are the biomes most often simulated for the future across all five GMCs refer to panel[c] for the ldquouncertaintyrdquo associated with the modal map) and (f) forest community typedistribution under the CGCM1 scenario among the warmest of future scenarios

well as northward along thewest coast of Canada intoAlaska (Figure 4)

The potential range of sev-eral subalpine conifers is sim-ulated to contract substantiallyin the western coterminousUnited States including En-glemann spruce (Picea engel-mannii) mountain hemlock(Tsuga mertensiana) and sev-eral species of true fir (Abies)Potential ranges for these sub-alpine species are simulated toexpand along the western coastof Canada and into AlaskaPotential future habitat for bigsagebrush (Artemisia triden-tata) an important shrub inthe inland West is largely ab-sent in the United States shift-ing into Canada This shrubis simulated to be replaced inthe United States by shrubschaparral and grasslands nowfound in arid regions of theSouthwest Potential habitatfor ponderosa pine (Pinusponderosa) is simulated to ex-pand in the western UnitedStates including on the WestCoast where many otherconifers are projected to con-tract

While the potential rangesof many taxa in the West shiftnorthward the topographiccomplexity of the region cre-ates less intuitive changes insome species distributions Po-tential habitat for some coniferspecies associated with mesicclimates shifts south and eastalong the Rocky Mountainswith for example forests typ-ical today of Glacier NationalPark becoming dominant inYellowstone National Park tothe southeast (Bartlein et al1997) The direction of changemay also differ from east towest within a speciesrsquo rangePotential habitat for paperbirch whose range spans thecontinent is simulated to con-tract northward in the eastern

772 BioScience September 2001 Vol 51 No 9

Articles

Figure 4 Simulated distributions and scenario agreement for Betula papyriferaPseudotsuga menziesii Pinus ponderosa and Artemisia tridentata Estimatedprobabilities of occurrence for a taxon simulated with observed modern climate (leftpanel) Comparison of the observed distributions with the simulated distributions underfuture climate conditions as generated by HADCM2S and CGCM1 for 2090ndash2099 (middlepanels) Gray indicates locations where the taxon is observed today and is simulated tooccur under future climate conditions red indicates locations where the taxon is observedtoday but is simulated to be absent under future climate conditions and blue indicateslocations where the taxon is absent today but is simulated to occur under future climateconditions Scenario agreement (right panel) Light purple indicates locations where thespecies is simulated to be present under the future climate of either the HADCM2S orCGCM1 scenario dark purple indicates locations where the species is simulated to bepresent under both future climate scenarios

United States but to expand southward along the RockyMountains in the West (Figure 4) The complex topographyof the West results in simulated future habitat for many treespecies that is disjunct Consequently dispersal to new habi-tats under climate change may be more difficult in the Westthan in the East where the future distributions of species aresimulated to be more continuous

Community richness Currie (2001) derived multipleregression models relating the broad-scale variability ofspecies richness of trees and terrestrial vertebrates acrossNorth America to the spatial patterns of summer and winterprecipitation and temperature These models were used to pre-dict patterns of species richness under five GCM scenariosThis approach assumes that richness will continue to covarywith climate in the future in the same way that it does today

The results indicate that contemporary patterns of richnesscorrelate strongly with temperature and less strongly with pre-cipitation These climate-richness relationships differ amongtaxonomic groups thus projected change in richness underclimate change also differed among groups Current treespecies richness is a positive function of temperature up to rel-atively high temperatures then is negatively related to furtherincreases in temperature Current tree species richness is alsoa positive function of precipitation Thus climatic warmingis predicted to lead to increased tree richness over most of thenorthern United States especially in the areas that are nowcoldest in the western mountains and near the CanadianBorder (Figure 5) Moderate decreases in richness (ndash20) arepredicted to occur in areas that are likely to experience dry-ing and very high temperatures such as the southwesterndeserts The greatest disagreement among the projectionswas for Pacific Northwest All models predict increases inrichness in that area but these increases may be modest (lt +50) to pronounced (gt +100)

Contemporary species richness of endotherms (birds andmammals) covaries strongly with temperature Richness inthese groups is maximal in moderately warm areas (thesouthern Appalachians and southern Rockies) and it de-creases in hotter areas This relationship may occur becauseambient heat serves as a direct energy subsidy for endothermsin cold climates but these organisms expend energy to dis-sipate heat in hot areas Endotherm richness is only weaklyrelated to precipitation Consequently under most climatechange scenarios endotherm richness is predicted to de-crease by over 25 in low-elevation areas in the SoutheastIncreases in richness (gt +11 to gt +100) are predicted forupper montane areas across the United States

Contemporary ectotherm (reptiles and amphibians) rich-ness is even more strongly related to temperature increasingmonotonically as temperature increases Ambient heat is alsoan energy subsidy for ectotherms even in the warmest areasof the earth Consequently climatic warming is predicted toincrease ectotherm richness over the entire coterminousUnited States For reptiles the increase is predicted to bemodest across the South and greater in the North Amphib-

ian richness in contrast also depends somewhat on precip-itation Because most GCMs predict that the southeasternUnited States will become somewhat drier in winter am-phibian richness in the Southeast is predicted to change lit-tle despite increased temperatures and to increase elsewhereDifferences among predictions of different GCMs lie mainlyin how dramatically richness is likely to increase

Threatened and endangered species How might thechanges in species richness predicted above influence en-dangered species hotspots (places where many endangeredspecies occur) We overlaid the maps of projected species rich-ness over maps of current hotspots for threatened and en-dangered species (Flather et al 1998) to determine the pro-portion of each endangerment hotspot area in the contiguousUnited States that is predicted to show an increase decreaseor no change in species richness for each taxonomic groupThe results indicated that reptiles and amphibians are expectedto increase in richness across all endangerment hotspots Onthe other hand bird and mammal richness may undergosignificant reductions in many endangerment hotspots es-pecially in the East (Figure 6)

Implications of assessment resultsIn contrast to public perceptions of large-scale forest lossunder global change the biome models project only a mod-est average loss of forest area (11) for the coterminousUnited States Lost forest is largely replaced by savanna andarid woodland biome types However the projected responseof forest habitats to global change scenarios is highly variableamong GCMs Forest habitats are projected to increase by 23under the HADCM2SUL model and are predicted to de-crease by 45 under the UKMO model This uncertaintypoints to cautious interpretation of our findings and to theneed for further research directed at understanding the fac-tors driving change in forest systems

Within the climate projections considered we did observesome common patterns in the response of forest communitytypes across the continent Areal expansion of habitats was pro-jected for oakndashhickory and oakndashpine in the East and ponderosapine and arid-tolerant hardwoods in the West The first threeof these are extremely valuable for forest products and ashabitat The heavy mast production of oakndashhickory for ex-ample is a food source for more than 180 different kinds ofvertebrates (Rogers 1990) The oak types also produce per-sistent coarse woody debris that benefits several ecologicalprocesses and organisms

Suitable habitats for several important community typeshowever are projected to greatly decrease in area or disappearfrom the coterminous United States These include alpine habi-tats subalpine sprucendashfir forests aspen the maplendashbeechndashbirchtype sagebrush and loblollyndashshortleaf pine communitiesSubalpine sprucendashfir has been decreasing in area becomingincreasingly fragmented and losing species richness sincethe last glacial period (Brown and Davis 1995) Other habi-tats such as sagebrush and aspen are being reduced in mod-

September 2001 Vol 51 No 9 BioScience 773

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

Land-use activities may appear to be too local in scale toaffect regional biome distributions However in some biomessuch as the North American Prairie land conversion is so ex-tensive that little native vegetation remains Land use canalso affect biomes to an extent greater than the sum of the areadirectly affected Conversion of natural vegetation to an-thropogenic cover types can alter the frequency and scale ofdisturbance agents that also define the character of biomesFor example fire suppression and grazing in the AmericanSouthwest are thought to be partially responsible for the in-vasion of woody vegetation into arid grassland habitats(Brown and Davis 1995) For a more compete discussion ofinteractions between climate change and disturbance seeDale et al (2001)

Feedbacks among climateland use and biodiversityClimate and land use often interact in ways that influence bio-diversity implying that these factors cannot be considered inisolation For example land use may modify climatic impactson species distributions by altering dispersal routes Whereland use creates barriers to dispersal for native species and fa-cilitates dispersal for exotic species climate change in human-dominated landscapes is likely to select for exotic species andagainst many native species (Malanson and Cairns 1997) Suchconstraints on dispersal are of concern especially around na-ture reserves Organisms ldquotrappedrdquo in reserves by surround-ing land use may become extinct if they are not able to dis-perse to increasingly suitable habitats in other nature reserves(Halpin 1997) Even in nature reserves ldquoweedyrdquo species willmost likely be quick to replace native species that succumb toclimate change

Climate and land use also jointly influence disturbancessuch as wildfire flooding and landslides Some land usespreset the landscape to be very sensitive to extreme climateevents leading to severe disturbance Logging can cause dry-ing of fuels and allow severe fires during normal drought pe-riods (Franklin and Forman 1987) This situation is thoughtto have transpired in Indonesia where slash-and-burn agri-cultural practices provided fuels and ignition sources whenan El Nintildeo event induced drought in 1997 Consequently vastareas of the rainforest burned possibly jeopardizing many en-demic species Roads and logging practices can similarly in-crease land sliding and flooding during storm events (Swan-son and Dryness 1975) Livestock grazing on the other handoften reduces fuel loads and reduces wildfire frequency andintensity for a given climate condition (Arno and Gruell1983)

Changes in land-cover patterns can also directly affect cli-mate which in turn influences biodiversity (Dale 1997) Forexample deforestation over large areas can cause reductionsin transpiration cloud formation and rainfall and increasedlevels of drying (Dickenson 1991) Such changes can lead todramatic shifts in biome type such as the replacement offorests by shrubs or grassland Similarly agricultural landuse and irrigation in the western Great Plains has been asso-

ciated with increased cloudiness and precipitation in theRocky Mountains of Colorado (Chase et al 1999)

Recent responses of biodiversity to global changeTrends in biodiversity to changes in climate and land useover the last 100 years provide a context for understanding fu-ture interactions During the past century average air tem-peratures have increased 0ndash2 ordmC over most of the UnitedStates and Canada (Watson et al 1998) Changes in precip-itation have been variable over this period increasing from5 to 20 over most of the United States but decreasing upto 20 in the Northern Rockies and California (Watson et al1998) The US population grew from 76 million in 1900 toover 270 million in 1998 Agricultural lands increased in areauntil 1900 then decreased slowly until 1950 they have re-mained stable since then (Maizel et al 1999) Between 1942and 1992 urban area increased 120 while protected areasincreased 80 (Flather et al 1999) Rural residential devel-opment has increased rapidly in the mountain west in recentdecades expanding human influences into semi-natural habi-tats

Many species and communities have responded to thesechanges in climate and land use For example forest declineand dieback are evident along the Atlantic and Pacific coastsand may be related to elevated CO2 levels and climate change(Mueller-Dombois 1992) The breeding ranges of some mo-bile species such as waterfowl have been expanding north-ward in association with climate amelioration (Abrahamand Jefferies 1997) Shifts in demography are apparent insome species Both amphibians and birds in Great Britain haveshifted breeding dates by 1 to 3 weeks earlier since the 1970sin association with increasing temperatures (Beebee 1995Crick et al 1997)

Species associated with human-dominated landscapeshave greatly expanded in recent years These include manylarge ungulate and small mammal species waterfowl andsome bird species associated with agricultural and urban en-vironments (Flather et al 1999) Some of these species are nowso abundant (eg deer) that the primary concern is controllingtheir populations Many exotic species have also become es-tablished in the United States and greatly expanded theirranges (Drake et al 1989)

At the same time several natural community types and nu-merous species have been greatly reduced by human activi-ties For example natural sprucendashfir longleaf pine andloblollyndashshortleaf pine forests now cover less than 2 oftheir presettlement ranges (Noss et al 1994) and are likely tobe further reduced under global warming Many species de-pendent upon these endangered and reduced ecosystems arecurrently in peril The number of species listed as threat-ened and endangered in the United States under the Endan-gered Species Act currently totals 1232 (USDI 2000) Factorscontributing to species endangerment include habitat con-version resource extraction and exotic species (Wilcove et al1998) The spatial distribution of such factors results in

September 2001 Vol 51 No 9 BioScience 767

Articles

species at risk being concentrated in particular regions ofthe United States especially the southern Appalachians thearid Southwest and coastal areas (Flather et al 1998)

Biodiversity under future global changeGiven past relationships among climate land use and bio-diversity how might biodiversity respond to future globalchange We assess potential vegetation response to projectedfuture climate change for doubled CO2 concentrations by us-ing the climate predictions of general circulation models(GCMs) (Table 1) as input to a set of different vegetation sim-ulation models The primary focus is on the potential conti-nental-scale response of forest vegetation as reflected inchanges in the distributions of biomes community types andtree species Although vegetation models incorporating land-use dynamics are under rapid development at local scales thecurrent state of knowledge does not allow for integrating theeffects of land use at the continental scale In our assessmentwe chose to emphasize forests as a key element of biodiver-sity which allowed us to draw implications for other organ-isms that require forest habitats We also simulate changes inspecies richness of trees and terrestrial vertebrates based onenergy theory (Currie 1991) and examine potential effects ofclimate change on locations where endangered species are con-centrated Results are for the coterminous United States un-less stated otherwise Each of these assessments is summarizedhere and reported in detail elsewhere (Table 2)

The models used to anticipate biodiversity responses tochanges in climate are based on empirical relationships be-tween organisms (vegetation and animals) and the environ-ments they currently occupy By using these relationships to

predict biodiversity response we are implicitly assuming thatthese environmentndashorganism relationships will remain un-altered in the future Although this approach can be criticizedfor not modeling some or all of the actual mechanisms lead-ing to shifts in vegetation and species ranges such an approachis commonly taken (Rogers and Randolph 2000) to initiallyassess ecological responses to global change scenarios It is alsoimportant to emphasize that the given GCMs are coarse-grid regionally smoothed outputs that do not allow depictionof local or even subregional climates Consequently the vegetation outputs modeled here must also be consideredcoarse and not sensitive to local phenomena

Climate change scenarios The GCMs differ in formu-lation hence predictions vary among the models Conse-quently the simulations are best considered as possible al-ternative views of the future with unknown likelihood ofoccurrence Both equilibrium and transient GCM scenariosare used in this assessment to incorporate a broad array of pos-sible futures (see Aber et al 2001) We put greater confidenceinto outcomes for which the models are in agreement we takedisagreement among the models as an indication of uncer-tainty Thus we report major findings for which most of themodels agree and we point out disagreement Still the un-certainty level for each climate and vegetation output is un-known and probably high given the uncertainty in fore-casting climate and subsequent vegetation responses

For the coterminous United States the different climate sce-narios all predict some level of warming and increased annualprecipitation (Table 1) Mean annual temperature increasesvary from 33degC to 58degC with the greatest warming at higherlatitudes Mean annual precipitation is predicted to increase

768 BioScience September 2001 Vol 51 No 9

Articles

Table 1 General circulation models and predictions for change from current average annual temperature (dT) andprecipitation (dP) for the coterminous United States under a doubling of atmospheric CO2 The equilibrium-type modelssimulate an instantaneous increase in CO2 and are run until equilibrium climate conditions emerge The newer transientmodels assume trace gases increase at 1 per year until 2100 and allow the climate to adjust while incorporatinginherent lags in the oceanndashatmosphere systems The HADCM2SUL and CGCM1 scenarios include the effects of sulfateaerosols The dT and dP values from the transient scenarios were calculated from averages of the last 30 years of thescenarios compared with the 1961ndash1990 means

Name Acronym Type Reference dT (oC) dP ()

Oregon State University OSU Equilibrium Schlesinger and Zhao 1989 3 21

Geophysical Fluids Dynamics Laborator y GFDLR30 Equilibrium Manabe et al 1990 42 189

Goddard Institute for Space Studies GISS Equilibrium Hansen et al 1988 44 51

United Kingdom Meteorological Office UKMO Equilibrium Wilson and Mitchell 1987 66 113

UKMO Hadley Centre HADCM2SUL Transient Johns et al 1997 28 229

UKMO Hadley Centre HADCM2GHG Transient Johns et al 1997 37 307

Canadian Climate Centre CGCM1 Transient Boer et al 2000 52 215

in the West with 20 to more than 50 increases in Cali-fornia Decreased precipitation of up to 30 is predicted forlocations in the Southeast Texas and the Northwest

Biomes The MAPSS biogeography model (Neilson 1995)projects biome response to climate as change in vegetationstructure and density based on light water and nutrient lim-itations (VEMAP members 1995 Bachelet et al 2001) Veg-etation is coupled directly to climate and hydrology andrules are applied to classify vegetation into biome types Themodel considers the effects of altered CO2 on plant physiol-ogy MAPSS simulates potential natural vegetation (specifi-cally life-forms) based on climate and does not include suc-cession or plant dispersal

The results project that potential forest area decreases byan average of 11 across the GCM scenarios with a range of+23 under the coolest scenarios and ndash45 under the hottestscenarios (Figure 2) Northeast mixed (hardwood and conifer)forests decrease by 72 in potential area on average (rangendash14 to ndash97) as they shift into Canada and increase in po-tential area continentally (Watson et al 1998) The potentialarea of eastern hardwoods decreases by an average of 34(range ndash93 to +51) These deciduous forests shift northreplacing northeastern mixed forests but are squeezed fromthe south by southeastern mixed forests or from the west bysavannas and grasslands depending on the scenario The po-tential range of southeastern mixed forests increases under allscenarios (average 37 range 25 to 57) while shiftingnorth This biome remains intact under cooler scenarios butis converted to savannas and grasslands in the South underthe hotter scenarios

Potential environments for alpine ecosystems all but dis-appear from the western mountains being overtaken by en-croaching forests Wet coniferous forests in the Northwest de-crease in potential area by 9 on average (range ndash54 to+21) The potential range of interior western pines change

little on average with a range of ndash30 to +39 The poten-tial range of shrublands and arid woodlands expand in the in-terior West and Great Plains encroaching on some grasslandsHowever grassland habitats may expand in the deserts ofthe Southwest parts of the Southeast and possibly in the up-per Midwest Thus the potential area of grassland could ei-ther decrease or increase

The projections agree on a single or on two vegetationclasses across 68 of the coterminous United States Locationsof greatest certainty are in the northern plains and FloridaRegions of great uncertainty are transition zones in the east-ern prairie and the West Taken in order of increasing tem-perature change the future scenarios imply that potentialforest range could expand with small amounts of warming butwould contract under the hotter scenarios

Forest community types and tree species Statisticalmodels are used to project the distributions of tree species withresults aggregated into community types These models pro-ject tree response to future climate based on current rela-tionships between trees and environmental variables such asclimate and soils Because physiological data are not required(as is the case for process models) many species can be mod-eled However these approaches do not include important for-est dynamics involving species interactions physiologicalprocesses such as CO2 uptake and tree dispersal and estab-lishment They also assume that speciesndashenvironment rela-tionships will remain the same under future climate condi-tions which may not be the case

Two statistical models are used The DISTRIB model ofIverson and Prasad (Iverson and Prasad 1998 Iverson et al1999a Prasad and Iverson 1999) was applied to the easternUnited States This model uses regression tree analysisbased on 33 environmental variables to predict the poten-tial future distribution of suitable habitat for 80 tree speciesAn index of species response (regional importance) was

September 2001 Vol 51 No 9 BioScience 769

Articles

Table 2 Biodiversity models used in this analysis

Response variables Model Model type Resolution Extent GCMs simulated Reference

Biomes community MAPSS Biogeographic 10 km grid coterminous HADCM2S HADCM2G Neilson 1995types processes United States CGCM1 OSU Bachelet et al

GFDLndashR30 GISS 2001UKMO

Tree species forest DISTRIB Statistical County Eastern United HADCM2S CGCM1 Iverson and community types regression tree States GFDLndashR30 GISS UKMO Prasad 1998

2001

Tree and shrub Response Statistical 25 km grid North America HADCM2S CGCM1 Shafer et al 2001species surface local regression

model

Species richness of Currie model Statistical 25deg x 25deg lat United States CGCM1 OSU GFDLndashR30 Currie 2001trees mammals birds and long south and Canada GISS UKMOreptiles amphibians of 50degN 5deg

long x 25deg lat north of 50degN

derived by multiplying the importance value (reflecting therelative abundance of a species in a community) by thearea for each county Community types were then definedby simply aggregating the importance values for individualtree species (Iverson and Prasad 2001) For western forestswe drew on the work of Shafer et al (2001) and Bartlein etal (1997) who use a local regression model to predict prob-ability of occurrence of tree species across North Americabased on climate and soils The results for dominant west-ern species are presented

In the eastern United States the area of habitat suitable foroakndashhickory expands in area following climate change by anaverage of 34 primarily to the north and east (Figure 3) Theoakndashpine habitat also expands by roughly 290 and is rep-resented throughout the Southeast On the other hand the

habitat capable of supporting the sprucendashfir and aspenndashbirchtypes are dramatically reduced (ndash97 and ndash92) and arelargely replaced by oakndashhickory and oakndashpine habitat Theloblollyndashshortleaf pine habitat is also reduced by 32 itshifts to the north and west while being replaced in its cur-rent zone by oakndashpine habitat The longleafndashslash pine habi-tat is reduced on average by 31 There is a higher level ofdisagreement among climate scenarios for the elmndashashndashcot-tonwood oakndashgumndashcypress and whitendashredndashjack pine typeswhich show increases in habitat under some scenarios and de-creases under others

These potential changes in habitat for community types re-flect the responses of individual tree species Seven of the 80species modeled were predicted to have their suitable habi-tat reduced in regional importance by at least 90 bigtooth

770 BioScience September 2001 Vol 51 No 9

Articles

Figure 2 Simulated potential distributions of biomes using the MAPSS biogeography model under six different GCMscenarios representing approximately 2 acute CO

2concentration (a) Color key for biomes (b) simulated current biome

distribution based on average (1961ndash1990) climate (c) uncertainty map (the number of unique biome types simulatedacross all six GCM scenarios are plotted) (d) biome distribution under the HADCM2SUL scenario among the coolest offuture warming scenarios (e) a modal map of future biome distributions (shown are the biomes most often simulated forthe future across all six GCMs refer to panel c for the ldquouncertaintyrdquo associated with the modal map) (f) biome distributionunder the CGCM1 scenario among the warmest of future scenarios

aspen (Populus grandiden-tata) quaking aspen (Ptremuloides ) sugar maple(Acer saccharum) northernwhite cedar (Thuja occiden-talis) balsam fir (Abies bal-samea) red pine (Pinusresinosa) and paper birch(Betula papyrifera) An ad-ditional 24 species would po-tentially decline by at least10 In contrast regionalimportance increased for 35species including 12 speciesthat increased suitable habi-tat by 100 or more in-cluding four species of oakand one hickory

Most speciesrsquo habitat wasprojected to move to thenorth 100 to 530 km for sev-eral species Some speciessuch as quaking aspen paperbirch northern white cedarbalsam fir and sugar maplehave the optimum latitudeof suitable habitat movenorth of the US border

All of the above analysesfor the eastern United Statesrelate to the potential distri-bution of suitable habitatnot actual distribution of thespecies The assumption isthat the species will get therethat there are no barriers orconstraints to migration(For newer projections thatconsider dispersal see Iver-son et al 1999b)

In the western UnitedStates the potential rangesof dominant rainforestconifers such as westernhemlock are projected to de-crease west of the CascadeMountains and expand intomountain ranges through-out the interior West Simu-lated potential habitat forDouglas fir (Pseudotsugamenziesii) also decreasesalong the western coast ofthe coterminous UnitedStates but expands east ofthe Cascades and Sierras as

September 2001 Vol 51 No 9 BioScience 771

Articles

Figure 3 Simulated potential distributions of forest community types in the eastern UnitedStates using the DISTRIB model under five different GMC scenarios representingapproximately 2 x CO

2concentration (a) Color key for forest types (b) current forest type

distribution based on 10000 actual forest inventory plots (c) uncertainty map (plotted arethe number of unique forest community types simulated across all five future GCM scenarios)(d) potential forest community type distribution under the HADCM2SUL scenario amongthe coolest of future warming scenarios (e) a modal map of future biome distributions(shown are the biomes most often simulated for the future across all five GMCs refer to panel[c] for the ldquouncertaintyrdquo associated with the modal map) and (f) forest community typedistribution under the CGCM1 scenario among the warmest of future scenarios

well as northward along thewest coast of Canada intoAlaska (Figure 4)

The potential range of sev-eral subalpine conifers is sim-ulated to contract substantiallyin the western coterminousUnited States including En-glemann spruce (Picea engel-mannii) mountain hemlock(Tsuga mertensiana) and sev-eral species of true fir (Abies)Potential ranges for these sub-alpine species are simulated toexpand along the western coastof Canada and into AlaskaPotential future habitat for bigsagebrush (Artemisia triden-tata) an important shrub inthe inland West is largely ab-sent in the United States shift-ing into Canada This shrubis simulated to be replaced inthe United States by shrubschaparral and grasslands nowfound in arid regions of theSouthwest Potential habitatfor ponderosa pine (Pinusponderosa) is simulated to ex-pand in the western UnitedStates including on the WestCoast where many otherconifers are projected to con-tract

While the potential rangesof many taxa in the West shiftnorthward the topographiccomplexity of the region cre-ates less intuitive changes insome species distributions Po-tential habitat for some coniferspecies associated with mesicclimates shifts south and eastalong the Rocky Mountainswith for example forests typ-ical today of Glacier NationalPark becoming dominant inYellowstone National Park tothe southeast (Bartlein et al1997) The direction of changemay also differ from east towest within a speciesrsquo rangePotential habitat for paperbirch whose range spans thecontinent is simulated to con-tract northward in the eastern

772 BioScience September 2001 Vol 51 No 9

Articles

Figure 4 Simulated distributions and scenario agreement for Betula papyriferaPseudotsuga menziesii Pinus ponderosa and Artemisia tridentata Estimatedprobabilities of occurrence for a taxon simulated with observed modern climate (leftpanel) Comparison of the observed distributions with the simulated distributions underfuture climate conditions as generated by HADCM2S and CGCM1 for 2090ndash2099 (middlepanels) Gray indicates locations where the taxon is observed today and is simulated tooccur under future climate conditions red indicates locations where the taxon is observedtoday but is simulated to be absent under future climate conditions and blue indicateslocations where the taxon is absent today but is simulated to occur under future climateconditions Scenario agreement (right panel) Light purple indicates locations where thespecies is simulated to be present under the future climate of either the HADCM2S orCGCM1 scenario dark purple indicates locations where the species is simulated to bepresent under both future climate scenarios

United States but to expand southward along the RockyMountains in the West (Figure 4) The complex topographyof the West results in simulated future habitat for many treespecies that is disjunct Consequently dispersal to new habi-tats under climate change may be more difficult in the Westthan in the East where the future distributions of species aresimulated to be more continuous

Community richness Currie (2001) derived multipleregression models relating the broad-scale variability ofspecies richness of trees and terrestrial vertebrates acrossNorth America to the spatial patterns of summer and winterprecipitation and temperature These models were used to pre-dict patterns of species richness under five GCM scenariosThis approach assumes that richness will continue to covarywith climate in the future in the same way that it does today

The results indicate that contemporary patterns of richnesscorrelate strongly with temperature and less strongly with pre-cipitation These climate-richness relationships differ amongtaxonomic groups thus projected change in richness underclimate change also differed among groups Current treespecies richness is a positive function of temperature up to rel-atively high temperatures then is negatively related to furtherincreases in temperature Current tree species richness is alsoa positive function of precipitation Thus climatic warmingis predicted to lead to increased tree richness over most of thenorthern United States especially in the areas that are nowcoldest in the western mountains and near the CanadianBorder (Figure 5) Moderate decreases in richness (ndash20) arepredicted to occur in areas that are likely to experience dry-ing and very high temperatures such as the southwesterndeserts The greatest disagreement among the projectionswas for Pacific Northwest All models predict increases inrichness in that area but these increases may be modest (lt +50) to pronounced (gt +100)

Contemporary species richness of endotherms (birds andmammals) covaries strongly with temperature Richness inthese groups is maximal in moderately warm areas (thesouthern Appalachians and southern Rockies) and it de-creases in hotter areas This relationship may occur becauseambient heat serves as a direct energy subsidy for endothermsin cold climates but these organisms expend energy to dis-sipate heat in hot areas Endotherm richness is only weaklyrelated to precipitation Consequently under most climatechange scenarios endotherm richness is predicted to de-crease by over 25 in low-elevation areas in the SoutheastIncreases in richness (gt +11 to gt +100) are predicted forupper montane areas across the United States

Contemporary ectotherm (reptiles and amphibians) rich-ness is even more strongly related to temperature increasingmonotonically as temperature increases Ambient heat is alsoan energy subsidy for ectotherms even in the warmest areasof the earth Consequently climatic warming is predicted toincrease ectotherm richness over the entire coterminousUnited States For reptiles the increase is predicted to bemodest across the South and greater in the North Amphib-

ian richness in contrast also depends somewhat on precip-itation Because most GCMs predict that the southeasternUnited States will become somewhat drier in winter am-phibian richness in the Southeast is predicted to change lit-tle despite increased temperatures and to increase elsewhereDifferences among predictions of different GCMs lie mainlyin how dramatically richness is likely to increase

Threatened and endangered species How might thechanges in species richness predicted above influence en-dangered species hotspots (places where many endangeredspecies occur) We overlaid the maps of projected species rich-ness over maps of current hotspots for threatened and en-dangered species (Flather et al 1998) to determine the pro-portion of each endangerment hotspot area in the contiguousUnited States that is predicted to show an increase decreaseor no change in species richness for each taxonomic groupThe results indicated that reptiles and amphibians are expectedto increase in richness across all endangerment hotspots Onthe other hand bird and mammal richness may undergosignificant reductions in many endangerment hotspots es-pecially in the East (Figure 6)

Implications of assessment resultsIn contrast to public perceptions of large-scale forest lossunder global change the biome models project only a mod-est average loss of forest area (11) for the coterminousUnited States Lost forest is largely replaced by savanna andarid woodland biome types However the projected responseof forest habitats to global change scenarios is highly variableamong GCMs Forest habitats are projected to increase by 23under the HADCM2SUL model and are predicted to de-crease by 45 under the UKMO model This uncertaintypoints to cautious interpretation of our findings and to theneed for further research directed at understanding the fac-tors driving change in forest systems

Within the climate projections considered we did observesome common patterns in the response of forest communitytypes across the continent Areal expansion of habitats was pro-jected for oakndashhickory and oakndashpine in the East and ponderosapine and arid-tolerant hardwoods in the West The first threeof these are extremely valuable for forest products and ashabitat The heavy mast production of oakndashhickory for ex-ample is a food source for more than 180 different kinds ofvertebrates (Rogers 1990) The oak types also produce per-sistent coarse woody debris that benefits several ecologicalprocesses and organisms

Suitable habitats for several important community typeshowever are projected to greatly decrease in area or disappearfrom the coterminous United States These include alpine habi-tats subalpine sprucendashfir forests aspen the maplendashbeechndashbirchtype sagebrush and loblollyndashshortleaf pine communitiesSubalpine sprucendashfir has been decreasing in area becomingincreasingly fragmented and losing species richness sincethe last glacial period (Brown and Davis 1995) Other habi-tats such as sagebrush and aspen are being reduced in mod-

September 2001 Vol 51 No 9 BioScience 773

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

species at risk being concentrated in particular regions ofthe United States especially the southern Appalachians thearid Southwest and coastal areas (Flather et al 1998)

Biodiversity under future global changeGiven past relationships among climate land use and bio-diversity how might biodiversity respond to future globalchange We assess potential vegetation response to projectedfuture climate change for doubled CO2 concentrations by us-ing the climate predictions of general circulation models(GCMs) (Table 1) as input to a set of different vegetation sim-ulation models The primary focus is on the potential conti-nental-scale response of forest vegetation as reflected inchanges in the distributions of biomes community types andtree species Although vegetation models incorporating land-use dynamics are under rapid development at local scales thecurrent state of knowledge does not allow for integrating theeffects of land use at the continental scale In our assessmentwe chose to emphasize forests as a key element of biodiver-sity which allowed us to draw implications for other organ-isms that require forest habitats We also simulate changes inspecies richness of trees and terrestrial vertebrates based onenergy theory (Currie 1991) and examine potential effects ofclimate change on locations where endangered species are con-centrated Results are for the coterminous United States un-less stated otherwise Each of these assessments is summarizedhere and reported in detail elsewhere (Table 2)

The models used to anticipate biodiversity responses tochanges in climate are based on empirical relationships be-tween organisms (vegetation and animals) and the environ-ments they currently occupy By using these relationships to

predict biodiversity response we are implicitly assuming thatthese environmentndashorganism relationships will remain un-altered in the future Although this approach can be criticizedfor not modeling some or all of the actual mechanisms lead-ing to shifts in vegetation and species ranges such an approachis commonly taken (Rogers and Randolph 2000) to initiallyassess ecological responses to global change scenarios It is alsoimportant to emphasize that the given GCMs are coarse-grid regionally smoothed outputs that do not allow depictionof local or even subregional climates Consequently the vegetation outputs modeled here must also be consideredcoarse and not sensitive to local phenomena

Climate change scenarios The GCMs differ in formu-lation hence predictions vary among the models Conse-quently the simulations are best considered as possible al-ternative views of the future with unknown likelihood ofoccurrence Both equilibrium and transient GCM scenariosare used in this assessment to incorporate a broad array of pos-sible futures (see Aber et al 2001) We put greater confidenceinto outcomes for which the models are in agreement we takedisagreement among the models as an indication of uncer-tainty Thus we report major findings for which most of themodels agree and we point out disagreement Still the un-certainty level for each climate and vegetation output is un-known and probably high given the uncertainty in fore-casting climate and subsequent vegetation responses

For the coterminous United States the different climate sce-narios all predict some level of warming and increased annualprecipitation (Table 1) Mean annual temperature increasesvary from 33degC to 58degC with the greatest warming at higherlatitudes Mean annual precipitation is predicted to increase

768 BioScience September 2001 Vol 51 No 9

Articles

Table 1 General circulation models and predictions for change from current average annual temperature (dT) andprecipitation (dP) for the coterminous United States under a doubling of atmospheric CO2 The equilibrium-type modelssimulate an instantaneous increase in CO2 and are run until equilibrium climate conditions emerge The newer transientmodels assume trace gases increase at 1 per year until 2100 and allow the climate to adjust while incorporatinginherent lags in the oceanndashatmosphere systems The HADCM2SUL and CGCM1 scenarios include the effects of sulfateaerosols The dT and dP values from the transient scenarios were calculated from averages of the last 30 years of thescenarios compared with the 1961ndash1990 means

Name Acronym Type Reference dT (oC) dP ()

Oregon State University OSU Equilibrium Schlesinger and Zhao 1989 3 21

Geophysical Fluids Dynamics Laborator y GFDLR30 Equilibrium Manabe et al 1990 42 189

Goddard Institute for Space Studies GISS Equilibrium Hansen et al 1988 44 51

United Kingdom Meteorological Office UKMO Equilibrium Wilson and Mitchell 1987 66 113

UKMO Hadley Centre HADCM2SUL Transient Johns et al 1997 28 229

UKMO Hadley Centre HADCM2GHG Transient Johns et al 1997 37 307

Canadian Climate Centre CGCM1 Transient Boer et al 2000 52 215

in the West with 20 to more than 50 increases in Cali-fornia Decreased precipitation of up to 30 is predicted forlocations in the Southeast Texas and the Northwest

Biomes The MAPSS biogeography model (Neilson 1995)projects biome response to climate as change in vegetationstructure and density based on light water and nutrient lim-itations (VEMAP members 1995 Bachelet et al 2001) Veg-etation is coupled directly to climate and hydrology andrules are applied to classify vegetation into biome types Themodel considers the effects of altered CO2 on plant physiol-ogy MAPSS simulates potential natural vegetation (specifi-cally life-forms) based on climate and does not include suc-cession or plant dispersal

The results project that potential forest area decreases byan average of 11 across the GCM scenarios with a range of+23 under the coolest scenarios and ndash45 under the hottestscenarios (Figure 2) Northeast mixed (hardwood and conifer)forests decrease by 72 in potential area on average (rangendash14 to ndash97) as they shift into Canada and increase in po-tential area continentally (Watson et al 1998) The potentialarea of eastern hardwoods decreases by an average of 34(range ndash93 to +51) These deciduous forests shift northreplacing northeastern mixed forests but are squeezed fromthe south by southeastern mixed forests or from the west bysavannas and grasslands depending on the scenario The po-tential range of southeastern mixed forests increases under allscenarios (average 37 range 25 to 57) while shiftingnorth This biome remains intact under cooler scenarios butis converted to savannas and grasslands in the South underthe hotter scenarios

Potential environments for alpine ecosystems all but dis-appear from the western mountains being overtaken by en-croaching forests Wet coniferous forests in the Northwest de-crease in potential area by 9 on average (range ndash54 to+21) The potential range of interior western pines change

little on average with a range of ndash30 to +39 The poten-tial range of shrublands and arid woodlands expand in the in-terior West and Great Plains encroaching on some grasslandsHowever grassland habitats may expand in the deserts ofthe Southwest parts of the Southeast and possibly in the up-per Midwest Thus the potential area of grassland could ei-ther decrease or increase

The projections agree on a single or on two vegetationclasses across 68 of the coterminous United States Locationsof greatest certainty are in the northern plains and FloridaRegions of great uncertainty are transition zones in the east-ern prairie and the West Taken in order of increasing tem-perature change the future scenarios imply that potentialforest range could expand with small amounts of warming butwould contract under the hotter scenarios

Forest community types and tree species Statisticalmodels are used to project the distributions of tree species withresults aggregated into community types These models pro-ject tree response to future climate based on current rela-tionships between trees and environmental variables such asclimate and soils Because physiological data are not required(as is the case for process models) many species can be mod-eled However these approaches do not include important for-est dynamics involving species interactions physiologicalprocesses such as CO2 uptake and tree dispersal and estab-lishment They also assume that speciesndashenvironment rela-tionships will remain the same under future climate condi-tions which may not be the case

Two statistical models are used The DISTRIB model ofIverson and Prasad (Iverson and Prasad 1998 Iverson et al1999a Prasad and Iverson 1999) was applied to the easternUnited States This model uses regression tree analysisbased on 33 environmental variables to predict the poten-tial future distribution of suitable habitat for 80 tree speciesAn index of species response (regional importance) was

September 2001 Vol 51 No 9 BioScience 769

Articles

Table 2 Biodiversity models used in this analysis

Response variables Model Model type Resolution Extent GCMs simulated Reference

Biomes community MAPSS Biogeographic 10 km grid coterminous HADCM2S HADCM2G Neilson 1995types processes United States CGCM1 OSU Bachelet et al

GFDLndashR30 GISS 2001UKMO

Tree species forest DISTRIB Statistical County Eastern United HADCM2S CGCM1 Iverson and community types regression tree States GFDLndashR30 GISS UKMO Prasad 1998

2001

Tree and shrub Response Statistical 25 km grid North America HADCM2S CGCM1 Shafer et al 2001species surface local regression

model

Species richness of Currie model Statistical 25deg x 25deg lat United States CGCM1 OSU GFDLndashR30 Currie 2001trees mammals birds and long south and Canada GISS UKMOreptiles amphibians of 50degN 5deg

long x 25deg lat north of 50degN

derived by multiplying the importance value (reflecting therelative abundance of a species in a community) by thearea for each county Community types were then definedby simply aggregating the importance values for individualtree species (Iverson and Prasad 2001) For western forestswe drew on the work of Shafer et al (2001) and Bartlein etal (1997) who use a local regression model to predict prob-ability of occurrence of tree species across North Americabased on climate and soils The results for dominant west-ern species are presented

In the eastern United States the area of habitat suitable foroakndashhickory expands in area following climate change by anaverage of 34 primarily to the north and east (Figure 3) Theoakndashpine habitat also expands by roughly 290 and is rep-resented throughout the Southeast On the other hand the

habitat capable of supporting the sprucendashfir and aspenndashbirchtypes are dramatically reduced (ndash97 and ndash92) and arelargely replaced by oakndashhickory and oakndashpine habitat Theloblollyndashshortleaf pine habitat is also reduced by 32 itshifts to the north and west while being replaced in its cur-rent zone by oakndashpine habitat The longleafndashslash pine habi-tat is reduced on average by 31 There is a higher level ofdisagreement among climate scenarios for the elmndashashndashcot-tonwood oakndashgumndashcypress and whitendashredndashjack pine typeswhich show increases in habitat under some scenarios and de-creases under others

These potential changes in habitat for community types re-flect the responses of individual tree species Seven of the 80species modeled were predicted to have their suitable habi-tat reduced in regional importance by at least 90 bigtooth

770 BioScience September 2001 Vol 51 No 9

Articles

Figure 2 Simulated potential distributions of biomes using the MAPSS biogeography model under six different GCMscenarios representing approximately 2 acute CO

2concentration (a) Color key for biomes (b) simulated current biome

distribution based on average (1961ndash1990) climate (c) uncertainty map (the number of unique biome types simulatedacross all six GCM scenarios are plotted) (d) biome distribution under the HADCM2SUL scenario among the coolest offuture warming scenarios (e) a modal map of future biome distributions (shown are the biomes most often simulated forthe future across all six GCMs refer to panel c for the ldquouncertaintyrdquo associated with the modal map) (f) biome distributionunder the CGCM1 scenario among the warmest of future scenarios

aspen (Populus grandiden-tata) quaking aspen (Ptremuloides ) sugar maple(Acer saccharum) northernwhite cedar (Thuja occiden-talis) balsam fir (Abies bal-samea) red pine (Pinusresinosa) and paper birch(Betula papyrifera) An ad-ditional 24 species would po-tentially decline by at least10 In contrast regionalimportance increased for 35species including 12 speciesthat increased suitable habi-tat by 100 or more in-cluding four species of oakand one hickory

Most speciesrsquo habitat wasprojected to move to thenorth 100 to 530 km for sev-eral species Some speciessuch as quaking aspen paperbirch northern white cedarbalsam fir and sugar maplehave the optimum latitudeof suitable habitat movenorth of the US border

All of the above analysesfor the eastern United Statesrelate to the potential distri-bution of suitable habitatnot actual distribution of thespecies The assumption isthat the species will get therethat there are no barriers orconstraints to migration(For newer projections thatconsider dispersal see Iver-son et al 1999b)

In the western UnitedStates the potential rangesof dominant rainforestconifers such as westernhemlock are projected to de-crease west of the CascadeMountains and expand intomountain ranges through-out the interior West Simu-lated potential habitat forDouglas fir (Pseudotsugamenziesii) also decreasesalong the western coast ofthe coterminous UnitedStates but expands east ofthe Cascades and Sierras as

September 2001 Vol 51 No 9 BioScience 771

Articles

Figure 3 Simulated potential distributions of forest community types in the eastern UnitedStates using the DISTRIB model under five different GMC scenarios representingapproximately 2 x CO

2concentration (a) Color key for forest types (b) current forest type

distribution based on 10000 actual forest inventory plots (c) uncertainty map (plotted arethe number of unique forest community types simulated across all five future GCM scenarios)(d) potential forest community type distribution under the HADCM2SUL scenario amongthe coolest of future warming scenarios (e) a modal map of future biome distributions(shown are the biomes most often simulated for the future across all five GMCs refer to panel[c] for the ldquouncertaintyrdquo associated with the modal map) and (f) forest community typedistribution under the CGCM1 scenario among the warmest of future scenarios

well as northward along thewest coast of Canada intoAlaska (Figure 4)

The potential range of sev-eral subalpine conifers is sim-ulated to contract substantiallyin the western coterminousUnited States including En-glemann spruce (Picea engel-mannii) mountain hemlock(Tsuga mertensiana) and sev-eral species of true fir (Abies)Potential ranges for these sub-alpine species are simulated toexpand along the western coastof Canada and into AlaskaPotential future habitat for bigsagebrush (Artemisia triden-tata) an important shrub inthe inland West is largely ab-sent in the United States shift-ing into Canada This shrubis simulated to be replaced inthe United States by shrubschaparral and grasslands nowfound in arid regions of theSouthwest Potential habitatfor ponderosa pine (Pinusponderosa) is simulated to ex-pand in the western UnitedStates including on the WestCoast where many otherconifers are projected to con-tract

While the potential rangesof many taxa in the West shiftnorthward the topographiccomplexity of the region cre-ates less intuitive changes insome species distributions Po-tential habitat for some coniferspecies associated with mesicclimates shifts south and eastalong the Rocky Mountainswith for example forests typ-ical today of Glacier NationalPark becoming dominant inYellowstone National Park tothe southeast (Bartlein et al1997) The direction of changemay also differ from east towest within a speciesrsquo rangePotential habitat for paperbirch whose range spans thecontinent is simulated to con-tract northward in the eastern

772 BioScience September 2001 Vol 51 No 9

Articles

Figure 4 Simulated distributions and scenario agreement for Betula papyriferaPseudotsuga menziesii Pinus ponderosa and Artemisia tridentata Estimatedprobabilities of occurrence for a taxon simulated with observed modern climate (leftpanel) Comparison of the observed distributions with the simulated distributions underfuture climate conditions as generated by HADCM2S and CGCM1 for 2090ndash2099 (middlepanels) Gray indicates locations where the taxon is observed today and is simulated tooccur under future climate conditions red indicates locations where the taxon is observedtoday but is simulated to be absent under future climate conditions and blue indicateslocations where the taxon is absent today but is simulated to occur under future climateconditions Scenario agreement (right panel) Light purple indicates locations where thespecies is simulated to be present under the future climate of either the HADCM2S orCGCM1 scenario dark purple indicates locations where the species is simulated to bepresent under both future climate scenarios

United States but to expand southward along the RockyMountains in the West (Figure 4) The complex topographyof the West results in simulated future habitat for many treespecies that is disjunct Consequently dispersal to new habi-tats under climate change may be more difficult in the Westthan in the East where the future distributions of species aresimulated to be more continuous

Community richness Currie (2001) derived multipleregression models relating the broad-scale variability ofspecies richness of trees and terrestrial vertebrates acrossNorth America to the spatial patterns of summer and winterprecipitation and temperature These models were used to pre-dict patterns of species richness under five GCM scenariosThis approach assumes that richness will continue to covarywith climate in the future in the same way that it does today

The results indicate that contemporary patterns of richnesscorrelate strongly with temperature and less strongly with pre-cipitation These climate-richness relationships differ amongtaxonomic groups thus projected change in richness underclimate change also differed among groups Current treespecies richness is a positive function of temperature up to rel-atively high temperatures then is negatively related to furtherincreases in temperature Current tree species richness is alsoa positive function of precipitation Thus climatic warmingis predicted to lead to increased tree richness over most of thenorthern United States especially in the areas that are nowcoldest in the western mountains and near the CanadianBorder (Figure 5) Moderate decreases in richness (ndash20) arepredicted to occur in areas that are likely to experience dry-ing and very high temperatures such as the southwesterndeserts The greatest disagreement among the projectionswas for Pacific Northwest All models predict increases inrichness in that area but these increases may be modest (lt +50) to pronounced (gt +100)

Contemporary species richness of endotherms (birds andmammals) covaries strongly with temperature Richness inthese groups is maximal in moderately warm areas (thesouthern Appalachians and southern Rockies) and it de-creases in hotter areas This relationship may occur becauseambient heat serves as a direct energy subsidy for endothermsin cold climates but these organisms expend energy to dis-sipate heat in hot areas Endotherm richness is only weaklyrelated to precipitation Consequently under most climatechange scenarios endotherm richness is predicted to de-crease by over 25 in low-elevation areas in the SoutheastIncreases in richness (gt +11 to gt +100) are predicted forupper montane areas across the United States

Contemporary ectotherm (reptiles and amphibians) rich-ness is even more strongly related to temperature increasingmonotonically as temperature increases Ambient heat is alsoan energy subsidy for ectotherms even in the warmest areasof the earth Consequently climatic warming is predicted toincrease ectotherm richness over the entire coterminousUnited States For reptiles the increase is predicted to bemodest across the South and greater in the North Amphib-

ian richness in contrast also depends somewhat on precip-itation Because most GCMs predict that the southeasternUnited States will become somewhat drier in winter am-phibian richness in the Southeast is predicted to change lit-tle despite increased temperatures and to increase elsewhereDifferences among predictions of different GCMs lie mainlyin how dramatically richness is likely to increase

Threatened and endangered species How might thechanges in species richness predicted above influence en-dangered species hotspots (places where many endangeredspecies occur) We overlaid the maps of projected species rich-ness over maps of current hotspots for threatened and en-dangered species (Flather et al 1998) to determine the pro-portion of each endangerment hotspot area in the contiguousUnited States that is predicted to show an increase decreaseor no change in species richness for each taxonomic groupThe results indicated that reptiles and amphibians are expectedto increase in richness across all endangerment hotspots Onthe other hand bird and mammal richness may undergosignificant reductions in many endangerment hotspots es-pecially in the East (Figure 6)

Implications of assessment resultsIn contrast to public perceptions of large-scale forest lossunder global change the biome models project only a mod-est average loss of forest area (11) for the coterminousUnited States Lost forest is largely replaced by savanna andarid woodland biome types However the projected responseof forest habitats to global change scenarios is highly variableamong GCMs Forest habitats are projected to increase by 23under the HADCM2SUL model and are predicted to de-crease by 45 under the UKMO model This uncertaintypoints to cautious interpretation of our findings and to theneed for further research directed at understanding the fac-tors driving change in forest systems

Within the climate projections considered we did observesome common patterns in the response of forest communitytypes across the continent Areal expansion of habitats was pro-jected for oakndashhickory and oakndashpine in the East and ponderosapine and arid-tolerant hardwoods in the West The first threeof these are extremely valuable for forest products and ashabitat The heavy mast production of oakndashhickory for ex-ample is a food source for more than 180 different kinds ofvertebrates (Rogers 1990) The oak types also produce per-sistent coarse woody debris that benefits several ecologicalprocesses and organisms

Suitable habitats for several important community typeshowever are projected to greatly decrease in area or disappearfrom the coterminous United States These include alpine habi-tats subalpine sprucendashfir forests aspen the maplendashbeechndashbirchtype sagebrush and loblollyndashshortleaf pine communitiesSubalpine sprucendashfir has been decreasing in area becomingincreasingly fragmented and losing species richness sincethe last glacial period (Brown and Davis 1995) Other habi-tats such as sagebrush and aspen are being reduced in mod-

September 2001 Vol 51 No 9 BioScience 773

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

in the West with 20 to more than 50 increases in Cali-fornia Decreased precipitation of up to 30 is predicted forlocations in the Southeast Texas and the Northwest

Biomes The MAPSS biogeography model (Neilson 1995)projects biome response to climate as change in vegetationstructure and density based on light water and nutrient lim-itations (VEMAP members 1995 Bachelet et al 2001) Veg-etation is coupled directly to climate and hydrology andrules are applied to classify vegetation into biome types Themodel considers the effects of altered CO2 on plant physiol-ogy MAPSS simulates potential natural vegetation (specifi-cally life-forms) based on climate and does not include suc-cession or plant dispersal

The results project that potential forest area decreases byan average of 11 across the GCM scenarios with a range of+23 under the coolest scenarios and ndash45 under the hottestscenarios (Figure 2) Northeast mixed (hardwood and conifer)forests decrease by 72 in potential area on average (rangendash14 to ndash97) as they shift into Canada and increase in po-tential area continentally (Watson et al 1998) The potentialarea of eastern hardwoods decreases by an average of 34(range ndash93 to +51) These deciduous forests shift northreplacing northeastern mixed forests but are squeezed fromthe south by southeastern mixed forests or from the west bysavannas and grasslands depending on the scenario The po-tential range of southeastern mixed forests increases under allscenarios (average 37 range 25 to 57) while shiftingnorth This biome remains intact under cooler scenarios butis converted to savannas and grasslands in the South underthe hotter scenarios

Potential environments for alpine ecosystems all but dis-appear from the western mountains being overtaken by en-croaching forests Wet coniferous forests in the Northwest de-crease in potential area by 9 on average (range ndash54 to+21) The potential range of interior western pines change

little on average with a range of ndash30 to +39 The poten-tial range of shrublands and arid woodlands expand in the in-terior West and Great Plains encroaching on some grasslandsHowever grassland habitats may expand in the deserts ofthe Southwest parts of the Southeast and possibly in the up-per Midwest Thus the potential area of grassland could ei-ther decrease or increase

The projections agree on a single or on two vegetationclasses across 68 of the coterminous United States Locationsof greatest certainty are in the northern plains and FloridaRegions of great uncertainty are transition zones in the east-ern prairie and the West Taken in order of increasing tem-perature change the future scenarios imply that potentialforest range could expand with small amounts of warming butwould contract under the hotter scenarios

Forest community types and tree species Statisticalmodels are used to project the distributions of tree species withresults aggregated into community types These models pro-ject tree response to future climate based on current rela-tionships between trees and environmental variables such asclimate and soils Because physiological data are not required(as is the case for process models) many species can be mod-eled However these approaches do not include important for-est dynamics involving species interactions physiologicalprocesses such as CO2 uptake and tree dispersal and estab-lishment They also assume that speciesndashenvironment rela-tionships will remain the same under future climate condi-tions which may not be the case

Two statistical models are used The DISTRIB model ofIverson and Prasad (Iverson and Prasad 1998 Iverson et al1999a Prasad and Iverson 1999) was applied to the easternUnited States This model uses regression tree analysisbased on 33 environmental variables to predict the poten-tial future distribution of suitable habitat for 80 tree speciesAn index of species response (regional importance) was

September 2001 Vol 51 No 9 BioScience 769

Articles

Table 2 Biodiversity models used in this analysis

Response variables Model Model type Resolution Extent GCMs simulated Reference

Biomes community MAPSS Biogeographic 10 km grid coterminous HADCM2S HADCM2G Neilson 1995types processes United States CGCM1 OSU Bachelet et al

GFDLndashR30 GISS 2001UKMO

Tree species forest DISTRIB Statistical County Eastern United HADCM2S CGCM1 Iverson and community types regression tree States GFDLndashR30 GISS UKMO Prasad 1998

2001

Tree and shrub Response Statistical 25 km grid North America HADCM2S CGCM1 Shafer et al 2001species surface local regression

model

Species richness of Currie model Statistical 25deg x 25deg lat United States CGCM1 OSU GFDLndashR30 Currie 2001trees mammals birds and long south and Canada GISS UKMOreptiles amphibians of 50degN 5deg

long x 25deg lat north of 50degN

derived by multiplying the importance value (reflecting therelative abundance of a species in a community) by thearea for each county Community types were then definedby simply aggregating the importance values for individualtree species (Iverson and Prasad 2001) For western forestswe drew on the work of Shafer et al (2001) and Bartlein etal (1997) who use a local regression model to predict prob-ability of occurrence of tree species across North Americabased on climate and soils The results for dominant west-ern species are presented

In the eastern United States the area of habitat suitable foroakndashhickory expands in area following climate change by anaverage of 34 primarily to the north and east (Figure 3) Theoakndashpine habitat also expands by roughly 290 and is rep-resented throughout the Southeast On the other hand the

habitat capable of supporting the sprucendashfir and aspenndashbirchtypes are dramatically reduced (ndash97 and ndash92) and arelargely replaced by oakndashhickory and oakndashpine habitat Theloblollyndashshortleaf pine habitat is also reduced by 32 itshifts to the north and west while being replaced in its cur-rent zone by oakndashpine habitat The longleafndashslash pine habi-tat is reduced on average by 31 There is a higher level ofdisagreement among climate scenarios for the elmndashashndashcot-tonwood oakndashgumndashcypress and whitendashredndashjack pine typeswhich show increases in habitat under some scenarios and de-creases under others

These potential changes in habitat for community types re-flect the responses of individual tree species Seven of the 80species modeled were predicted to have their suitable habi-tat reduced in regional importance by at least 90 bigtooth

770 BioScience September 2001 Vol 51 No 9

Articles

Figure 2 Simulated potential distributions of biomes using the MAPSS biogeography model under six different GCMscenarios representing approximately 2 acute CO

2concentration (a) Color key for biomes (b) simulated current biome

distribution based on average (1961ndash1990) climate (c) uncertainty map (the number of unique biome types simulatedacross all six GCM scenarios are plotted) (d) biome distribution under the HADCM2SUL scenario among the coolest offuture warming scenarios (e) a modal map of future biome distributions (shown are the biomes most often simulated forthe future across all six GCMs refer to panel c for the ldquouncertaintyrdquo associated with the modal map) (f) biome distributionunder the CGCM1 scenario among the warmest of future scenarios

aspen (Populus grandiden-tata) quaking aspen (Ptremuloides ) sugar maple(Acer saccharum) northernwhite cedar (Thuja occiden-talis) balsam fir (Abies bal-samea) red pine (Pinusresinosa) and paper birch(Betula papyrifera) An ad-ditional 24 species would po-tentially decline by at least10 In contrast regionalimportance increased for 35species including 12 speciesthat increased suitable habi-tat by 100 or more in-cluding four species of oakand one hickory

Most speciesrsquo habitat wasprojected to move to thenorth 100 to 530 km for sev-eral species Some speciessuch as quaking aspen paperbirch northern white cedarbalsam fir and sugar maplehave the optimum latitudeof suitable habitat movenorth of the US border

All of the above analysesfor the eastern United Statesrelate to the potential distri-bution of suitable habitatnot actual distribution of thespecies The assumption isthat the species will get therethat there are no barriers orconstraints to migration(For newer projections thatconsider dispersal see Iver-son et al 1999b)

In the western UnitedStates the potential rangesof dominant rainforestconifers such as westernhemlock are projected to de-crease west of the CascadeMountains and expand intomountain ranges through-out the interior West Simu-lated potential habitat forDouglas fir (Pseudotsugamenziesii) also decreasesalong the western coast ofthe coterminous UnitedStates but expands east ofthe Cascades and Sierras as

September 2001 Vol 51 No 9 BioScience 771

Articles

Figure 3 Simulated potential distributions of forest community types in the eastern UnitedStates using the DISTRIB model under five different GMC scenarios representingapproximately 2 x CO

2concentration (a) Color key for forest types (b) current forest type

distribution based on 10000 actual forest inventory plots (c) uncertainty map (plotted arethe number of unique forest community types simulated across all five future GCM scenarios)(d) potential forest community type distribution under the HADCM2SUL scenario amongthe coolest of future warming scenarios (e) a modal map of future biome distributions(shown are the biomes most often simulated for the future across all five GMCs refer to panel[c] for the ldquouncertaintyrdquo associated with the modal map) and (f) forest community typedistribution under the CGCM1 scenario among the warmest of future scenarios

well as northward along thewest coast of Canada intoAlaska (Figure 4)

The potential range of sev-eral subalpine conifers is sim-ulated to contract substantiallyin the western coterminousUnited States including En-glemann spruce (Picea engel-mannii) mountain hemlock(Tsuga mertensiana) and sev-eral species of true fir (Abies)Potential ranges for these sub-alpine species are simulated toexpand along the western coastof Canada and into AlaskaPotential future habitat for bigsagebrush (Artemisia triden-tata) an important shrub inthe inland West is largely ab-sent in the United States shift-ing into Canada This shrubis simulated to be replaced inthe United States by shrubschaparral and grasslands nowfound in arid regions of theSouthwest Potential habitatfor ponderosa pine (Pinusponderosa) is simulated to ex-pand in the western UnitedStates including on the WestCoast where many otherconifers are projected to con-tract

While the potential rangesof many taxa in the West shiftnorthward the topographiccomplexity of the region cre-ates less intuitive changes insome species distributions Po-tential habitat for some coniferspecies associated with mesicclimates shifts south and eastalong the Rocky Mountainswith for example forests typ-ical today of Glacier NationalPark becoming dominant inYellowstone National Park tothe southeast (Bartlein et al1997) The direction of changemay also differ from east towest within a speciesrsquo rangePotential habitat for paperbirch whose range spans thecontinent is simulated to con-tract northward in the eastern

772 BioScience September 2001 Vol 51 No 9

Articles

Figure 4 Simulated distributions and scenario agreement for Betula papyriferaPseudotsuga menziesii Pinus ponderosa and Artemisia tridentata Estimatedprobabilities of occurrence for a taxon simulated with observed modern climate (leftpanel) Comparison of the observed distributions with the simulated distributions underfuture climate conditions as generated by HADCM2S and CGCM1 for 2090ndash2099 (middlepanels) Gray indicates locations where the taxon is observed today and is simulated tooccur under future climate conditions red indicates locations where the taxon is observedtoday but is simulated to be absent under future climate conditions and blue indicateslocations where the taxon is absent today but is simulated to occur under future climateconditions Scenario agreement (right panel) Light purple indicates locations where thespecies is simulated to be present under the future climate of either the HADCM2S orCGCM1 scenario dark purple indicates locations where the species is simulated to bepresent under both future climate scenarios

United States but to expand southward along the RockyMountains in the West (Figure 4) The complex topographyof the West results in simulated future habitat for many treespecies that is disjunct Consequently dispersal to new habi-tats under climate change may be more difficult in the Westthan in the East where the future distributions of species aresimulated to be more continuous

Community richness Currie (2001) derived multipleregression models relating the broad-scale variability ofspecies richness of trees and terrestrial vertebrates acrossNorth America to the spatial patterns of summer and winterprecipitation and temperature These models were used to pre-dict patterns of species richness under five GCM scenariosThis approach assumes that richness will continue to covarywith climate in the future in the same way that it does today

The results indicate that contemporary patterns of richnesscorrelate strongly with temperature and less strongly with pre-cipitation These climate-richness relationships differ amongtaxonomic groups thus projected change in richness underclimate change also differed among groups Current treespecies richness is a positive function of temperature up to rel-atively high temperatures then is negatively related to furtherincreases in temperature Current tree species richness is alsoa positive function of precipitation Thus climatic warmingis predicted to lead to increased tree richness over most of thenorthern United States especially in the areas that are nowcoldest in the western mountains and near the CanadianBorder (Figure 5) Moderate decreases in richness (ndash20) arepredicted to occur in areas that are likely to experience dry-ing and very high temperatures such as the southwesterndeserts The greatest disagreement among the projectionswas for Pacific Northwest All models predict increases inrichness in that area but these increases may be modest (lt +50) to pronounced (gt +100)

Contemporary species richness of endotherms (birds andmammals) covaries strongly with temperature Richness inthese groups is maximal in moderately warm areas (thesouthern Appalachians and southern Rockies) and it de-creases in hotter areas This relationship may occur becauseambient heat serves as a direct energy subsidy for endothermsin cold climates but these organisms expend energy to dis-sipate heat in hot areas Endotherm richness is only weaklyrelated to precipitation Consequently under most climatechange scenarios endotherm richness is predicted to de-crease by over 25 in low-elevation areas in the SoutheastIncreases in richness (gt +11 to gt +100) are predicted forupper montane areas across the United States

Contemporary ectotherm (reptiles and amphibians) rich-ness is even more strongly related to temperature increasingmonotonically as temperature increases Ambient heat is alsoan energy subsidy for ectotherms even in the warmest areasof the earth Consequently climatic warming is predicted toincrease ectotherm richness over the entire coterminousUnited States For reptiles the increase is predicted to bemodest across the South and greater in the North Amphib-

ian richness in contrast also depends somewhat on precip-itation Because most GCMs predict that the southeasternUnited States will become somewhat drier in winter am-phibian richness in the Southeast is predicted to change lit-tle despite increased temperatures and to increase elsewhereDifferences among predictions of different GCMs lie mainlyin how dramatically richness is likely to increase

Threatened and endangered species How might thechanges in species richness predicted above influence en-dangered species hotspots (places where many endangeredspecies occur) We overlaid the maps of projected species rich-ness over maps of current hotspots for threatened and en-dangered species (Flather et al 1998) to determine the pro-portion of each endangerment hotspot area in the contiguousUnited States that is predicted to show an increase decreaseor no change in species richness for each taxonomic groupThe results indicated that reptiles and amphibians are expectedto increase in richness across all endangerment hotspots Onthe other hand bird and mammal richness may undergosignificant reductions in many endangerment hotspots es-pecially in the East (Figure 6)

Implications of assessment resultsIn contrast to public perceptions of large-scale forest lossunder global change the biome models project only a mod-est average loss of forest area (11) for the coterminousUnited States Lost forest is largely replaced by savanna andarid woodland biome types However the projected responseof forest habitats to global change scenarios is highly variableamong GCMs Forest habitats are projected to increase by 23under the HADCM2SUL model and are predicted to de-crease by 45 under the UKMO model This uncertaintypoints to cautious interpretation of our findings and to theneed for further research directed at understanding the fac-tors driving change in forest systems

Within the climate projections considered we did observesome common patterns in the response of forest communitytypes across the continent Areal expansion of habitats was pro-jected for oakndashhickory and oakndashpine in the East and ponderosapine and arid-tolerant hardwoods in the West The first threeof these are extremely valuable for forest products and ashabitat The heavy mast production of oakndashhickory for ex-ample is a food source for more than 180 different kinds ofvertebrates (Rogers 1990) The oak types also produce per-sistent coarse woody debris that benefits several ecologicalprocesses and organisms

Suitable habitats for several important community typeshowever are projected to greatly decrease in area or disappearfrom the coterminous United States These include alpine habi-tats subalpine sprucendashfir forests aspen the maplendashbeechndashbirchtype sagebrush and loblollyndashshortleaf pine communitiesSubalpine sprucendashfir has been decreasing in area becomingincreasingly fragmented and losing species richness sincethe last glacial period (Brown and Davis 1995) Other habi-tats such as sagebrush and aspen are being reduced in mod-

September 2001 Vol 51 No 9 BioScience 773

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

derived by multiplying the importance value (reflecting therelative abundance of a species in a community) by thearea for each county Community types were then definedby simply aggregating the importance values for individualtree species (Iverson and Prasad 2001) For western forestswe drew on the work of Shafer et al (2001) and Bartlein etal (1997) who use a local regression model to predict prob-ability of occurrence of tree species across North Americabased on climate and soils The results for dominant west-ern species are presented

In the eastern United States the area of habitat suitable foroakndashhickory expands in area following climate change by anaverage of 34 primarily to the north and east (Figure 3) Theoakndashpine habitat also expands by roughly 290 and is rep-resented throughout the Southeast On the other hand the

habitat capable of supporting the sprucendashfir and aspenndashbirchtypes are dramatically reduced (ndash97 and ndash92) and arelargely replaced by oakndashhickory and oakndashpine habitat Theloblollyndashshortleaf pine habitat is also reduced by 32 itshifts to the north and west while being replaced in its cur-rent zone by oakndashpine habitat The longleafndashslash pine habi-tat is reduced on average by 31 There is a higher level ofdisagreement among climate scenarios for the elmndashashndashcot-tonwood oakndashgumndashcypress and whitendashredndashjack pine typeswhich show increases in habitat under some scenarios and de-creases under others

These potential changes in habitat for community types re-flect the responses of individual tree species Seven of the 80species modeled were predicted to have their suitable habi-tat reduced in regional importance by at least 90 bigtooth

770 BioScience September 2001 Vol 51 No 9

Articles

Figure 2 Simulated potential distributions of biomes using the MAPSS biogeography model under six different GCMscenarios representing approximately 2 acute CO

2concentration (a) Color key for biomes (b) simulated current biome

distribution based on average (1961ndash1990) climate (c) uncertainty map (the number of unique biome types simulatedacross all six GCM scenarios are plotted) (d) biome distribution under the HADCM2SUL scenario among the coolest offuture warming scenarios (e) a modal map of future biome distributions (shown are the biomes most often simulated forthe future across all six GCMs refer to panel c for the ldquouncertaintyrdquo associated with the modal map) (f) biome distributionunder the CGCM1 scenario among the warmest of future scenarios

aspen (Populus grandiden-tata) quaking aspen (Ptremuloides ) sugar maple(Acer saccharum) northernwhite cedar (Thuja occiden-talis) balsam fir (Abies bal-samea) red pine (Pinusresinosa) and paper birch(Betula papyrifera) An ad-ditional 24 species would po-tentially decline by at least10 In contrast regionalimportance increased for 35species including 12 speciesthat increased suitable habi-tat by 100 or more in-cluding four species of oakand one hickory

Most speciesrsquo habitat wasprojected to move to thenorth 100 to 530 km for sev-eral species Some speciessuch as quaking aspen paperbirch northern white cedarbalsam fir and sugar maplehave the optimum latitudeof suitable habitat movenorth of the US border

All of the above analysesfor the eastern United Statesrelate to the potential distri-bution of suitable habitatnot actual distribution of thespecies The assumption isthat the species will get therethat there are no barriers orconstraints to migration(For newer projections thatconsider dispersal see Iver-son et al 1999b)

In the western UnitedStates the potential rangesof dominant rainforestconifers such as westernhemlock are projected to de-crease west of the CascadeMountains and expand intomountain ranges through-out the interior West Simu-lated potential habitat forDouglas fir (Pseudotsugamenziesii) also decreasesalong the western coast ofthe coterminous UnitedStates but expands east ofthe Cascades and Sierras as

September 2001 Vol 51 No 9 BioScience 771

Articles

Figure 3 Simulated potential distributions of forest community types in the eastern UnitedStates using the DISTRIB model under five different GMC scenarios representingapproximately 2 x CO

2concentration (a) Color key for forest types (b) current forest type

distribution based on 10000 actual forest inventory plots (c) uncertainty map (plotted arethe number of unique forest community types simulated across all five future GCM scenarios)(d) potential forest community type distribution under the HADCM2SUL scenario amongthe coolest of future warming scenarios (e) a modal map of future biome distributions(shown are the biomes most often simulated for the future across all five GMCs refer to panel[c] for the ldquouncertaintyrdquo associated with the modal map) and (f) forest community typedistribution under the CGCM1 scenario among the warmest of future scenarios

well as northward along thewest coast of Canada intoAlaska (Figure 4)

The potential range of sev-eral subalpine conifers is sim-ulated to contract substantiallyin the western coterminousUnited States including En-glemann spruce (Picea engel-mannii) mountain hemlock(Tsuga mertensiana) and sev-eral species of true fir (Abies)Potential ranges for these sub-alpine species are simulated toexpand along the western coastof Canada and into AlaskaPotential future habitat for bigsagebrush (Artemisia triden-tata) an important shrub inthe inland West is largely ab-sent in the United States shift-ing into Canada This shrubis simulated to be replaced inthe United States by shrubschaparral and grasslands nowfound in arid regions of theSouthwest Potential habitatfor ponderosa pine (Pinusponderosa) is simulated to ex-pand in the western UnitedStates including on the WestCoast where many otherconifers are projected to con-tract

While the potential rangesof many taxa in the West shiftnorthward the topographiccomplexity of the region cre-ates less intuitive changes insome species distributions Po-tential habitat for some coniferspecies associated with mesicclimates shifts south and eastalong the Rocky Mountainswith for example forests typ-ical today of Glacier NationalPark becoming dominant inYellowstone National Park tothe southeast (Bartlein et al1997) The direction of changemay also differ from east towest within a speciesrsquo rangePotential habitat for paperbirch whose range spans thecontinent is simulated to con-tract northward in the eastern

772 BioScience September 2001 Vol 51 No 9

Articles

Figure 4 Simulated distributions and scenario agreement for Betula papyriferaPseudotsuga menziesii Pinus ponderosa and Artemisia tridentata Estimatedprobabilities of occurrence for a taxon simulated with observed modern climate (leftpanel) Comparison of the observed distributions with the simulated distributions underfuture climate conditions as generated by HADCM2S and CGCM1 for 2090ndash2099 (middlepanels) Gray indicates locations where the taxon is observed today and is simulated tooccur under future climate conditions red indicates locations where the taxon is observedtoday but is simulated to be absent under future climate conditions and blue indicateslocations where the taxon is absent today but is simulated to occur under future climateconditions Scenario agreement (right panel) Light purple indicates locations where thespecies is simulated to be present under the future climate of either the HADCM2S orCGCM1 scenario dark purple indicates locations where the species is simulated to bepresent under both future climate scenarios

United States but to expand southward along the RockyMountains in the West (Figure 4) The complex topographyof the West results in simulated future habitat for many treespecies that is disjunct Consequently dispersal to new habi-tats under climate change may be more difficult in the Westthan in the East where the future distributions of species aresimulated to be more continuous

Community richness Currie (2001) derived multipleregression models relating the broad-scale variability ofspecies richness of trees and terrestrial vertebrates acrossNorth America to the spatial patterns of summer and winterprecipitation and temperature These models were used to pre-dict patterns of species richness under five GCM scenariosThis approach assumes that richness will continue to covarywith climate in the future in the same way that it does today

The results indicate that contemporary patterns of richnesscorrelate strongly with temperature and less strongly with pre-cipitation These climate-richness relationships differ amongtaxonomic groups thus projected change in richness underclimate change also differed among groups Current treespecies richness is a positive function of temperature up to rel-atively high temperatures then is negatively related to furtherincreases in temperature Current tree species richness is alsoa positive function of precipitation Thus climatic warmingis predicted to lead to increased tree richness over most of thenorthern United States especially in the areas that are nowcoldest in the western mountains and near the CanadianBorder (Figure 5) Moderate decreases in richness (ndash20) arepredicted to occur in areas that are likely to experience dry-ing and very high temperatures such as the southwesterndeserts The greatest disagreement among the projectionswas for Pacific Northwest All models predict increases inrichness in that area but these increases may be modest (lt +50) to pronounced (gt +100)

Contemporary species richness of endotherms (birds andmammals) covaries strongly with temperature Richness inthese groups is maximal in moderately warm areas (thesouthern Appalachians and southern Rockies) and it de-creases in hotter areas This relationship may occur becauseambient heat serves as a direct energy subsidy for endothermsin cold climates but these organisms expend energy to dis-sipate heat in hot areas Endotherm richness is only weaklyrelated to precipitation Consequently under most climatechange scenarios endotherm richness is predicted to de-crease by over 25 in low-elevation areas in the SoutheastIncreases in richness (gt +11 to gt +100) are predicted forupper montane areas across the United States

Contemporary ectotherm (reptiles and amphibians) rich-ness is even more strongly related to temperature increasingmonotonically as temperature increases Ambient heat is alsoan energy subsidy for ectotherms even in the warmest areasof the earth Consequently climatic warming is predicted toincrease ectotherm richness over the entire coterminousUnited States For reptiles the increase is predicted to bemodest across the South and greater in the North Amphib-

ian richness in contrast also depends somewhat on precip-itation Because most GCMs predict that the southeasternUnited States will become somewhat drier in winter am-phibian richness in the Southeast is predicted to change lit-tle despite increased temperatures and to increase elsewhereDifferences among predictions of different GCMs lie mainlyin how dramatically richness is likely to increase

Threatened and endangered species How might thechanges in species richness predicted above influence en-dangered species hotspots (places where many endangeredspecies occur) We overlaid the maps of projected species rich-ness over maps of current hotspots for threatened and en-dangered species (Flather et al 1998) to determine the pro-portion of each endangerment hotspot area in the contiguousUnited States that is predicted to show an increase decreaseor no change in species richness for each taxonomic groupThe results indicated that reptiles and amphibians are expectedto increase in richness across all endangerment hotspots Onthe other hand bird and mammal richness may undergosignificant reductions in many endangerment hotspots es-pecially in the East (Figure 6)

Implications of assessment resultsIn contrast to public perceptions of large-scale forest lossunder global change the biome models project only a mod-est average loss of forest area (11) for the coterminousUnited States Lost forest is largely replaced by savanna andarid woodland biome types However the projected responseof forest habitats to global change scenarios is highly variableamong GCMs Forest habitats are projected to increase by 23under the HADCM2SUL model and are predicted to de-crease by 45 under the UKMO model This uncertaintypoints to cautious interpretation of our findings and to theneed for further research directed at understanding the fac-tors driving change in forest systems

Within the climate projections considered we did observesome common patterns in the response of forest communitytypes across the continent Areal expansion of habitats was pro-jected for oakndashhickory and oakndashpine in the East and ponderosapine and arid-tolerant hardwoods in the West The first threeof these are extremely valuable for forest products and ashabitat The heavy mast production of oakndashhickory for ex-ample is a food source for more than 180 different kinds ofvertebrates (Rogers 1990) The oak types also produce per-sistent coarse woody debris that benefits several ecologicalprocesses and organisms

Suitable habitats for several important community typeshowever are projected to greatly decrease in area or disappearfrom the coterminous United States These include alpine habi-tats subalpine sprucendashfir forests aspen the maplendashbeechndashbirchtype sagebrush and loblollyndashshortleaf pine communitiesSubalpine sprucendashfir has been decreasing in area becomingincreasingly fragmented and losing species richness sincethe last glacial period (Brown and Davis 1995) Other habi-tats such as sagebrush and aspen are being reduced in mod-

September 2001 Vol 51 No 9 BioScience 773

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

aspen (Populus grandiden-tata) quaking aspen (Ptremuloides ) sugar maple(Acer saccharum) northernwhite cedar (Thuja occiden-talis) balsam fir (Abies bal-samea) red pine (Pinusresinosa) and paper birch(Betula papyrifera) An ad-ditional 24 species would po-tentially decline by at least10 In contrast regionalimportance increased for 35species including 12 speciesthat increased suitable habi-tat by 100 or more in-cluding four species of oakand one hickory

Most speciesrsquo habitat wasprojected to move to thenorth 100 to 530 km for sev-eral species Some speciessuch as quaking aspen paperbirch northern white cedarbalsam fir and sugar maplehave the optimum latitudeof suitable habitat movenorth of the US border

All of the above analysesfor the eastern United Statesrelate to the potential distri-bution of suitable habitatnot actual distribution of thespecies The assumption isthat the species will get therethat there are no barriers orconstraints to migration(For newer projections thatconsider dispersal see Iver-son et al 1999b)

In the western UnitedStates the potential rangesof dominant rainforestconifers such as westernhemlock are projected to de-crease west of the CascadeMountains and expand intomountain ranges through-out the interior West Simu-lated potential habitat forDouglas fir (Pseudotsugamenziesii) also decreasesalong the western coast ofthe coterminous UnitedStates but expands east ofthe Cascades and Sierras as

September 2001 Vol 51 No 9 BioScience 771

Articles

Figure 3 Simulated potential distributions of forest community types in the eastern UnitedStates using the DISTRIB model under five different GMC scenarios representingapproximately 2 x CO

2concentration (a) Color key for forest types (b) current forest type

distribution based on 10000 actual forest inventory plots (c) uncertainty map (plotted arethe number of unique forest community types simulated across all five future GCM scenarios)(d) potential forest community type distribution under the HADCM2SUL scenario amongthe coolest of future warming scenarios (e) a modal map of future biome distributions(shown are the biomes most often simulated for the future across all five GMCs refer to panel[c] for the ldquouncertaintyrdquo associated with the modal map) and (f) forest community typedistribution under the CGCM1 scenario among the warmest of future scenarios

well as northward along thewest coast of Canada intoAlaska (Figure 4)

The potential range of sev-eral subalpine conifers is sim-ulated to contract substantiallyin the western coterminousUnited States including En-glemann spruce (Picea engel-mannii) mountain hemlock(Tsuga mertensiana) and sev-eral species of true fir (Abies)Potential ranges for these sub-alpine species are simulated toexpand along the western coastof Canada and into AlaskaPotential future habitat for bigsagebrush (Artemisia triden-tata) an important shrub inthe inland West is largely ab-sent in the United States shift-ing into Canada This shrubis simulated to be replaced inthe United States by shrubschaparral and grasslands nowfound in arid regions of theSouthwest Potential habitatfor ponderosa pine (Pinusponderosa) is simulated to ex-pand in the western UnitedStates including on the WestCoast where many otherconifers are projected to con-tract

While the potential rangesof many taxa in the West shiftnorthward the topographiccomplexity of the region cre-ates less intuitive changes insome species distributions Po-tential habitat for some coniferspecies associated with mesicclimates shifts south and eastalong the Rocky Mountainswith for example forests typ-ical today of Glacier NationalPark becoming dominant inYellowstone National Park tothe southeast (Bartlein et al1997) The direction of changemay also differ from east towest within a speciesrsquo rangePotential habitat for paperbirch whose range spans thecontinent is simulated to con-tract northward in the eastern

772 BioScience September 2001 Vol 51 No 9

Articles

Figure 4 Simulated distributions and scenario agreement for Betula papyriferaPseudotsuga menziesii Pinus ponderosa and Artemisia tridentata Estimatedprobabilities of occurrence for a taxon simulated with observed modern climate (leftpanel) Comparison of the observed distributions with the simulated distributions underfuture climate conditions as generated by HADCM2S and CGCM1 for 2090ndash2099 (middlepanels) Gray indicates locations where the taxon is observed today and is simulated tooccur under future climate conditions red indicates locations where the taxon is observedtoday but is simulated to be absent under future climate conditions and blue indicateslocations where the taxon is absent today but is simulated to occur under future climateconditions Scenario agreement (right panel) Light purple indicates locations where thespecies is simulated to be present under the future climate of either the HADCM2S orCGCM1 scenario dark purple indicates locations where the species is simulated to bepresent under both future climate scenarios

United States but to expand southward along the RockyMountains in the West (Figure 4) The complex topographyof the West results in simulated future habitat for many treespecies that is disjunct Consequently dispersal to new habi-tats under climate change may be more difficult in the Westthan in the East where the future distributions of species aresimulated to be more continuous

Community richness Currie (2001) derived multipleregression models relating the broad-scale variability ofspecies richness of trees and terrestrial vertebrates acrossNorth America to the spatial patterns of summer and winterprecipitation and temperature These models were used to pre-dict patterns of species richness under five GCM scenariosThis approach assumes that richness will continue to covarywith climate in the future in the same way that it does today

The results indicate that contemporary patterns of richnesscorrelate strongly with temperature and less strongly with pre-cipitation These climate-richness relationships differ amongtaxonomic groups thus projected change in richness underclimate change also differed among groups Current treespecies richness is a positive function of temperature up to rel-atively high temperatures then is negatively related to furtherincreases in temperature Current tree species richness is alsoa positive function of precipitation Thus climatic warmingis predicted to lead to increased tree richness over most of thenorthern United States especially in the areas that are nowcoldest in the western mountains and near the CanadianBorder (Figure 5) Moderate decreases in richness (ndash20) arepredicted to occur in areas that are likely to experience dry-ing and very high temperatures such as the southwesterndeserts The greatest disagreement among the projectionswas for Pacific Northwest All models predict increases inrichness in that area but these increases may be modest (lt +50) to pronounced (gt +100)

Contemporary species richness of endotherms (birds andmammals) covaries strongly with temperature Richness inthese groups is maximal in moderately warm areas (thesouthern Appalachians and southern Rockies) and it de-creases in hotter areas This relationship may occur becauseambient heat serves as a direct energy subsidy for endothermsin cold climates but these organisms expend energy to dis-sipate heat in hot areas Endotherm richness is only weaklyrelated to precipitation Consequently under most climatechange scenarios endotherm richness is predicted to de-crease by over 25 in low-elevation areas in the SoutheastIncreases in richness (gt +11 to gt +100) are predicted forupper montane areas across the United States

Contemporary ectotherm (reptiles and amphibians) rich-ness is even more strongly related to temperature increasingmonotonically as temperature increases Ambient heat is alsoan energy subsidy for ectotherms even in the warmest areasof the earth Consequently climatic warming is predicted toincrease ectotherm richness over the entire coterminousUnited States For reptiles the increase is predicted to bemodest across the South and greater in the North Amphib-

ian richness in contrast also depends somewhat on precip-itation Because most GCMs predict that the southeasternUnited States will become somewhat drier in winter am-phibian richness in the Southeast is predicted to change lit-tle despite increased temperatures and to increase elsewhereDifferences among predictions of different GCMs lie mainlyin how dramatically richness is likely to increase

Threatened and endangered species How might thechanges in species richness predicted above influence en-dangered species hotspots (places where many endangeredspecies occur) We overlaid the maps of projected species rich-ness over maps of current hotspots for threatened and en-dangered species (Flather et al 1998) to determine the pro-portion of each endangerment hotspot area in the contiguousUnited States that is predicted to show an increase decreaseor no change in species richness for each taxonomic groupThe results indicated that reptiles and amphibians are expectedto increase in richness across all endangerment hotspots Onthe other hand bird and mammal richness may undergosignificant reductions in many endangerment hotspots es-pecially in the East (Figure 6)

Implications of assessment resultsIn contrast to public perceptions of large-scale forest lossunder global change the biome models project only a mod-est average loss of forest area (11) for the coterminousUnited States Lost forest is largely replaced by savanna andarid woodland biome types However the projected responseof forest habitats to global change scenarios is highly variableamong GCMs Forest habitats are projected to increase by 23under the HADCM2SUL model and are predicted to de-crease by 45 under the UKMO model This uncertaintypoints to cautious interpretation of our findings and to theneed for further research directed at understanding the fac-tors driving change in forest systems

Within the climate projections considered we did observesome common patterns in the response of forest communitytypes across the continent Areal expansion of habitats was pro-jected for oakndashhickory and oakndashpine in the East and ponderosapine and arid-tolerant hardwoods in the West The first threeof these are extremely valuable for forest products and ashabitat The heavy mast production of oakndashhickory for ex-ample is a food source for more than 180 different kinds ofvertebrates (Rogers 1990) The oak types also produce per-sistent coarse woody debris that benefits several ecologicalprocesses and organisms

Suitable habitats for several important community typeshowever are projected to greatly decrease in area or disappearfrom the coterminous United States These include alpine habi-tats subalpine sprucendashfir forests aspen the maplendashbeechndashbirchtype sagebrush and loblollyndashshortleaf pine communitiesSubalpine sprucendashfir has been decreasing in area becomingincreasingly fragmented and losing species richness sincethe last glacial period (Brown and Davis 1995) Other habi-tats such as sagebrush and aspen are being reduced in mod-

September 2001 Vol 51 No 9 BioScience 773

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

well as northward along thewest coast of Canada intoAlaska (Figure 4)

The potential range of sev-eral subalpine conifers is sim-ulated to contract substantiallyin the western coterminousUnited States including En-glemann spruce (Picea engel-mannii) mountain hemlock(Tsuga mertensiana) and sev-eral species of true fir (Abies)Potential ranges for these sub-alpine species are simulated toexpand along the western coastof Canada and into AlaskaPotential future habitat for bigsagebrush (Artemisia triden-tata) an important shrub inthe inland West is largely ab-sent in the United States shift-ing into Canada This shrubis simulated to be replaced inthe United States by shrubschaparral and grasslands nowfound in arid regions of theSouthwest Potential habitatfor ponderosa pine (Pinusponderosa) is simulated to ex-pand in the western UnitedStates including on the WestCoast where many otherconifers are projected to con-tract

While the potential rangesof many taxa in the West shiftnorthward the topographiccomplexity of the region cre-ates less intuitive changes insome species distributions Po-tential habitat for some coniferspecies associated with mesicclimates shifts south and eastalong the Rocky Mountainswith for example forests typ-ical today of Glacier NationalPark becoming dominant inYellowstone National Park tothe southeast (Bartlein et al1997) The direction of changemay also differ from east towest within a speciesrsquo rangePotential habitat for paperbirch whose range spans thecontinent is simulated to con-tract northward in the eastern

772 BioScience September 2001 Vol 51 No 9

Articles

Figure 4 Simulated distributions and scenario agreement for Betula papyriferaPseudotsuga menziesii Pinus ponderosa and Artemisia tridentata Estimatedprobabilities of occurrence for a taxon simulated with observed modern climate (leftpanel) Comparison of the observed distributions with the simulated distributions underfuture climate conditions as generated by HADCM2S and CGCM1 for 2090ndash2099 (middlepanels) Gray indicates locations where the taxon is observed today and is simulated tooccur under future climate conditions red indicates locations where the taxon is observedtoday but is simulated to be absent under future climate conditions and blue indicateslocations where the taxon is absent today but is simulated to occur under future climateconditions Scenario agreement (right panel) Light purple indicates locations where thespecies is simulated to be present under the future climate of either the HADCM2S orCGCM1 scenario dark purple indicates locations where the species is simulated to bepresent under both future climate scenarios

United States but to expand southward along the RockyMountains in the West (Figure 4) The complex topographyof the West results in simulated future habitat for many treespecies that is disjunct Consequently dispersal to new habi-tats under climate change may be more difficult in the Westthan in the East where the future distributions of species aresimulated to be more continuous

Community richness Currie (2001) derived multipleregression models relating the broad-scale variability ofspecies richness of trees and terrestrial vertebrates acrossNorth America to the spatial patterns of summer and winterprecipitation and temperature These models were used to pre-dict patterns of species richness under five GCM scenariosThis approach assumes that richness will continue to covarywith climate in the future in the same way that it does today

The results indicate that contemporary patterns of richnesscorrelate strongly with temperature and less strongly with pre-cipitation These climate-richness relationships differ amongtaxonomic groups thus projected change in richness underclimate change also differed among groups Current treespecies richness is a positive function of temperature up to rel-atively high temperatures then is negatively related to furtherincreases in temperature Current tree species richness is alsoa positive function of precipitation Thus climatic warmingis predicted to lead to increased tree richness over most of thenorthern United States especially in the areas that are nowcoldest in the western mountains and near the CanadianBorder (Figure 5) Moderate decreases in richness (ndash20) arepredicted to occur in areas that are likely to experience dry-ing and very high temperatures such as the southwesterndeserts The greatest disagreement among the projectionswas for Pacific Northwest All models predict increases inrichness in that area but these increases may be modest (lt +50) to pronounced (gt +100)

Contemporary species richness of endotherms (birds andmammals) covaries strongly with temperature Richness inthese groups is maximal in moderately warm areas (thesouthern Appalachians and southern Rockies) and it de-creases in hotter areas This relationship may occur becauseambient heat serves as a direct energy subsidy for endothermsin cold climates but these organisms expend energy to dis-sipate heat in hot areas Endotherm richness is only weaklyrelated to precipitation Consequently under most climatechange scenarios endotherm richness is predicted to de-crease by over 25 in low-elevation areas in the SoutheastIncreases in richness (gt +11 to gt +100) are predicted forupper montane areas across the United States

Contemporary ectotherm (reptiles and amphibians) rich-ness is even more strongly related to temperature increasingmonotonically as temperature increases Ambient heat is alsoan energy subsidy for ectotherms even in the warmest areasof the earth Consequently climatic warming is predicted toincrease ectotherm richness over the entire coterminousUnited States For reptiles the increase is predicted to bemodest across the South and greater in the North Amphib-

ian richness in contrast also depends somewhat on precip-itation Because most GCMs predict that the southeasternUnited States will become somewhat drier in winter am-phibian richness in the Southeast is predicted to change lit-tle despite increased temperatures and to increase elsewhereDifferences among predictions of different GCMs lie mainlyin how dramatically richness is likely to increase

Threatened and endangered species How might thechanges in species richness predicted above influence en-dangered species hotspots (places where many endangeredspecies occur) We overlaid the maps of projected species rich-ness over maps of current hotspots for threatened and en-dangered species (Flather et al 1998) to determine the pro-portion of each endangerment hotspot area in the contiguousUnited States that is predicted to show an increase decreaseor no change in species richness for each taxonomic groupThe results indicated that reptiles and amphibians are expectedto increase in richness across all endangerment hotspots Onthe other hand bird and mammal richness may undergosignificant reductions in many endangerment hotspots es-pecially in the East (Figure 6)

Implications of assessment resultsIn contrast to public perceptions of large-scale forest lossunder global change the biome models project only a mod-est average loss of forest area (11) for the coterminousUnited States Lost forest is largely replaced by savanna andarid woodland biome types However the projected responseof forest habitats to global change scenarios is highly variableamong GCMs Forest habitats are projected to increase by 23under the HADCM2SUL model and are predicted to de-crease by 45 under the UKMO model This uncertaintypoints to cautious interpretation of our findings and to theneed for further research directed at understanding the fac-tors driving change in forest systems

Within the climate projections considered we did observesome common patterns in the response of forest communitytypes across the continent Areal expansion of habitats was pro-jected for oakndashhickory and oakndashpine in the East and ponderosapine and arid-tolerant hardwoods in the West The first threeof these are extremely valuable for forest products and ashabitat The heavy mast production of oakndashhickory for ex-ample is a food source for more than 180 different kinds ofvertebrates (Rogers 1990) The oak types also produce per-sistent coarse woody debris that benefits several ecologicalprocesses and organisms

Suitable habitats for several important community typeshowever are projected to greatly decrease in area or disappearfrom the coterminous United States These include alpine habi-tats subalpine sprucendashfir forests aspen the maplendashbeechndashbirchtype sagebrush and loblollyndashshortleaf pine communitiesSubalpine sprucendashfir has been decreasing in area becomingincreasingly fragmented and losing species richness sincethe last glacial period (Brown and Davis 1995) Other habi-tats such as sagebrush and aspen are being reduced in mod-

September 2001 Vol 51 No 9 BioScience 773

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

United States but to expand southward along the RockyMountains in the West (Figure 4) The complex topographyof the West results in simulated future habitat for many treespecies that is disjunct Consequently dispersal to new habi-tats under climate change may be more difficult in the Westthan in the East where the future distributions of species aresimulated to be more continuous

Community richness Currie (2001) derived multipleregression models relating the broad-scale variability ofspecies richness of trees and terrestrial vertebrates acrossNorth America to the spatial patterns of summer and winterprecipitation and temperature These models were used to pre-dict patterns of species richness under five GCM scenariosThis approach assumes that richness will continue to covarywith climate in the future in the same way that it does today

The results indicate that contemporary patterns of richnesscorrelate strongly with temperature and less strongly with pre-cipitation These climate-richness relationships differ amongtaxonomic groups thus projected change in richness underclimate change also differed among groups Current treespecies richness is a positive function of temperature up to rel-atively high temperatures then is negatively related to furtherincreases in temperature Current tree species richness is alsoa positive function of precipitation Thus climatic warmingis predicted to lead to increased tree richness over most of thenorthern United States especially in the areas that are nowcoldest in the western mountains and near the CanadianBorder (Figure 5) Moderate decreases in richness (ndash20) arepredicted to occur in areas that are likely to experience dry-ing and very high temperatures such as the southwesterndeserts The greatest disagreement among the projectionswas for Pacific Northwest All models predict increases inrichness in that area but these increases may be modest (lt +50) to pronounced (gt +100)

Contemporary species richness of endotherms (birds andmammals) covaries strongly with temperature Richness inthese groups is maximal in moderately warm areas (thesouthern Appalachians and southern Rockies) and it de-creases in hotter areas This relationship may occur becauseambient heat serves as a direct energy subsidy for endothermsin cold climates but these organisms expend energy to dis-sipate heat in hot areas Endotherm richness is only weaklyrelated to precipitation Consequently under most climatechange scenarios endotherm richness is predicted to de-crease by over 25 in low-elevation areas in the SoutheastIncreases in richness (gt +11 to gt +100) are predicted forupper montane areas across the United States

Contemporary ectotherm (reptiles and amphibians) rich-ness is even more strongly related to temperature increasingmonotonically as temperature increases Ambient heat is alsoan energy subsidy for ectotherms even in the warmest areasof the earth Consequently climatic warming is predicted toincrease ectotherm richness over the entire coterminousUnited States For reptiles the increase is predicted to bemodest across the South and greater in the North Amphib-

ian richness in contrast also depends somewhat on precip-itation Because most GCMs predict that the southeasternUnited States will become somewhat drier in winter am-phibian richness in the Southeast is predicted to change lit-tle despite increased temperatures and to increase elsewhereDifferences among predictions of different GCMs lie mainlyin how dramatically richness is likely to increase

Threatened and endangered species How might thechanges in species richness predicted above influence en-dangered species hotspots (places where many endangeredspecies occur) We overlaid the maps of projected species rich-ness over maps of current hotspots for threatened and en-dangered species (Flather et al 1998) to determine the pro-portion of each endangerment hotspot area in the contiguousUnited States that is predicted to show an increase decreaseor no change in species richness for each taxonomic groupThe results indicated that reptiles and amphibians are expectedto increase in richness across all endangerment hotspots Onthe other hand bird and mammal richness may undergosignificant reductions in many endangerment hotspots es-pecially in the East (Figure 6)

Implications of assessment resultsIn contrast to public perceptions of large-scale forest lossunder global change the biome models project only a mod-est average loss of forest area (11) for the coterminousUnited States Lost forest is largely replaced by savanna andarid woodland biome types However the projected responseof forest habitats to global change scenarios is highly variableamong GCMs Forest habitats are projected to increase by 23under the HADCM2SUL model and are predicted to de-crease by 45 under the UKMO model This uncertaintypoints to cautious interpretation of our findings and to theneed for further research directed at understanding the fac-tors driving change in forest systems

Within the climate projections considered we did observesome common patterns in the response of forest communitytypes across the continent Areal expansion of habitats was pro-jected for oakndashhickory and oakndashpine in the East and ponderosapine and arid-tolerant hardwoods in the West The first threeof these are extremely valuable for forest products and ashabitat The heavy mast production of oakndashhickory for ex-ample is a food source for more than 180 different kinds ofvertebrates (Rogers 1990) The oak types also produce per-sistent coarse woody debris that benefits several ecologicalprocesses and organisms

Suitable habitats for several important community typeshowever are projected to greatly decrease in area or disappearfrom the coterminous United States These include alpine habi-tats subalpine sprucendashfir forests aspen the maplendashbeechndashbirchtype sagebrush and loblollyndashshortleaf pine communitiesSubalpine sprucendashfir has been decreasing in area becomingincreasingly fragmented and losing species richness sincethe last glacial period (Brown and Davis 1995) Other habi-tats such as sagebrush and aspen are being reduced in mod-

September 2001 Vol 51 No 9 BioScience 773

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

ern times by conifer encroachment grazing and exurbandevelopment Climate warming is projected to further reducethese community types and the multitude of species depen-dent upon them Quaking aspen for example is a keystonespecies in the west where it is often the only abundant hard-wood among the conifer forests Its palatability soft wood andrapid decomposition result in many species of plants and an-imals being dependent upon it Subalpine sprucendashfir nowsupports several species that require wilderness conditions(eg wolverine Gulo luscus) Sagebrush has several obligatespecies of vertebrates The maplendashbeechndashbirch type includes

several important tree species such as redmaple (Acer rubrus) sugar maple blackcherry (Prunus serotina) American beech(Fagus grandifolia) and yellow birch (Betulaalleghaniensis) Reductions of these specieswill influence many associated organisms

Suitable habitats for the forest types pro-jected to decrease in the coterminousUnited States are generally expected to in-crease in Canada Some of these northerlylocations are underlain by permafrost andhave nutrient-poor soils More work isneeded to determine which species couldtolerate future conditions in the northernecosystems Important policy questionsarise from the cross-border migration ofspecies and ecosystems that are increas-ingly threatened in the United States but in-creasingly common in Canada Suchchanges might suggest that the current na-tional regulations and incentives on biodi-versity be supplemented with internationalregulations and incentives

Our projections of species richness basedon energy theory suggest that climatechange will favor greater tree species rich-ness over much of the coterminous UnitedStates Climatic conditions are also pro-jected to become more favorable for am-phibians and reptiles Whether these cli-mate changes will counter the currentdecline in amphibians caused by pollutionultraviolet radiation land use and otherfactors will require further study Climateconditions are predicted to lead to lowerbird and mammal richness across thesouthern United States What are the im-plications of this projection for the manybird species that winter in the southernUnited States and breed to the north Per-haps of greatest concern are currentlythreatened or endangered bird and mam-mal species that are restricted to endan-germent hotspots in the Southeast Theprojected losses of species richness here

could further imperil these populations Individual speciesstudies will be needed to begin to understand the implicationsof these changes

These projections for species richness raise interestingquestions for management If climate change increases the po-tential for amphibian and reptile species richness whichspecies are likely to disperse to the newly suitable locations(Figure 7) Are there opportunities to introduce desirablespecies to the newly suitable habitats and select against non-desirable species that are good dispersers The projections forbirds and mammals indicate no change or slight increases in

774 BioScience September 2001 Vol 51 No 9

Articles

Figure 5 Predicted changes in species richness for trees (top left panel) birds(middle left) and amphibians (bottom left) after climate change relative tocurrent species richness The predictions represent the mean richness predictedby five different GCMs Maps of uncertainty (on the right) represent the extent ofdisagreement among GCMs The number of different classes of richness (amongthose used in the figures on the left) predicted by the five different GCMs isshown Thus in areas represented in blue (one class) all GCMs lead to the samepredicted change in richness whereas in ocher-colored areas three or more of theGCMs predicted changes that fell in different classes

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

richness in the North but decreases in the South Whichspecies in the southern areas are most likely to go extinct Canmanagement strategies be used to buffer these losses

Limitations and caveatsThe projections described above reflect the assumptions andformulations of our models The results indicate potentialtrends in biodiversity if environmentndashorganism relation-ships remain unchanged in the future and if organisms do notencounter constraints that limit their ability to track climatechanges It is important to keep in mind the factors that maymodify the relationship among climate and biomes com-munities and species

Our approach was to project change in suitable habitats forvarious levels of biodiversity under the climate projections ofseveral GCMs Again we emphasize that the accuracy of theclimate projections is unknown Our biodiversity modelsalso add an unknown level of uncertainty to our projectionsof habitat change Also the climate and biodiversity modelswere done on relatively coarse spatial resolutions Local vari-ation in climate due to topography or other factors could re-sult in species or communities being able to persist in suitablemicrohabitats even though the coarser-resolution modelsproject no suitable habitats in the location Another as-sumption of our approach is that the current locations ofspecies communities and biomes reflect environmental con-ditions suitable for all life-history stages of these organisms

In reality however the habitat requirements for seedling es-tablishment may not be identical to those for the survival andreproduction of adult trees and shrubs

The likelihood that organisms will be able to disperse tonewly suitable habitats will vary considerably among speciesThus differential rates of species dispersal will be a key de-terminant of future biodiversity patterns During theHolocene dominant plant species migrated at rates that al-lowed them to keep pace with climate change (Davis 1989)However the estimated rates of dispersal during the Holoceneabout 10ndash45 km per century (Davis and Zabinski 1992)were much slower than the potential geographic shifts impliedby our analysis of forest communities It is unclear how fastorganisms may disperse in modern landscapes subjected tovarious levels and types of land uses It is likely that disper-sal rates will be slower than the rate of climate change and thatweedy species will be better able than many other species todisperse through human-dominated landscapes

These differential rates of dispersal suggest that actualplant communities under climate change will not resemblethose predicted based on communityndashenvironment rela-tionships Instead the new communities will initially be dom-inated by the subset of species (especially weeds) that arebest able to track climate change Complex vegetation dy-namics should then follow as better competitors gradually ar-rive at these sites Adding to the complexity are the dynam-ics of the species that occupied the site under previous climatic

September 2001 Vol 51 No 9 BioScience 775

Articles

Figure 6 Proportion of hotspot areas that are predicted to lose (blue) gain (orange) or show no change (green) in speciesrichness by taxonomic group The numbers below hotspot names indicate the number of threatened and endangered speciesoccurring in that hotspot The numbers below the pie charts refer to the proportion of species for a given taxon (as definedby the row) in the hotspot (defined by the column) The results indicated that reptiles and amphibians are expected toincrease in richness across all endangerment hotspots hence data are not shown for these groups

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

conditions While some species are likely to vacate unsuitablesites rather quickly others may persist for decades to centuriesFranklin et al (1992) suggested that adult trees in old-growthforests in the Pacific Northwest may persist long after chang-ing climate no longer allows regeneration These forests couldthen change quickly as senescence or disturbance clears thesite allowing emigrating species to establish

A final caveat on the projections is that they do not con-sider land use disturbance or fine-scaled vegetation dy-namics (see Dale et al 2001) Some community types suchas prairie and coastal chaparral are now dominated by intensehuman land use which has altered the expected pattern of dis-turbance and negatively influenced native species Of the 63factors that have contributed to species endangerment in theUnited States land-use intensification associated with resi-dential and urban development forest management grazingand environmental contamination are the most commonfactors cited in nine out of the 12 endangerment hotspots(Flather et al 1998) As the population of the United Statesincreases the area of seminatural habitats will be further re-duced causing deviations from the relationship between cli-mate and biodiversity

This list of limitations makes evident the complexity of theinteraction of global change and biodiversity Although it isimportant for researchers to attempt to predict future globalchange and consequences for biodiversity society needs to beaware that the accuracy of these predictions may be mixedThe complex suite of interactions initiated by climate changecould well lead to some unexpected outcomes How to copewith such surprises may be one of the greatest challenges offuture global change

Coping strategiesManaging global climate-change impacts on biodiversity in-volves avoidance of impending climate and land-use changesaltering those changes or accepting the changes and dealingwith their impacts Strategies to slow global change includereducing the atmospheric concentration of greenhouse gaseshuman-induced disturbances and land-cover changes Con-trol of greenhouse gas emissions requires reduced use of fos-sil fuels and less harvesting of large trees for short-turnoverproducts as well as the establishment of new locations ormeans of carbon storage by such actions as planting large ar-eas with rapidly growing trees or enhancing their carbon se-questration potential However carbon storage in large plan-tations of monocultures of nonnative species would jeopardizenative diversity if those species replace natural vegetationStrategies for reducing changes in land cover and use in-clude management of human population growth land-useplanning and land-use regulation and incentives programsThese strategies can be designed to foster biodiversity if thatgoal is included in the overall plan

Where the maintenance of ecosystem processes and nativespecies is a priority effective strategies may differ with loca-tion community type and management objective For com-munities that are unlikely to migrate to suitable environ-ments elsewhere (eg subalpine and alpine communities) itmay be appropriate to minimize change by manipulatingvegetation structure composition or disturbance regimesto favor the current community For communities that maybe able to reach newly suitable habitats a reasonable strategymay be to manage some of the current habitat as a reservoiruntil the community is reestablished in the new locationsOther portions of the current habitat may be managed to en-courage change to the new species and communities more ap-propriate for the new environment Global change could of-fer opportunities to restore communities that are nowdegraded In this case management to induce rapid changemay allow for the establishment of species deemed to be de-sirable by society

In some cases diversity can be preserved only through re-serves set aside to protect species in the face of global landchanges Halpin (1997) offers a strategic framework of this typefor nature reserves The framework involves both maintain-ing current communities and facilitating natural dispersal oforganisms across elevational and latitudinal gradients (egvia migration or dispersal corridors) The five categories ofmanagement prescriptions presented by Halpin are (1) se-lection of redundant reserves (2) selection of reserves that pro-vide habitat diversity (3) management for buffer-zone flex-ibility (4) management for landscape connectivity and (5)management for habitat maintenance The exact prescriptionwill vary depending on characteristics of the species andcommunities to be preserved and their habitats For thespecies most at risk seed banks and captive breeding and rear-ing approaches may be necessary until new suitable habitatsdevelop

776 BioScience September 2001 Vol 51 No 9

Articles

Figure 7 Current trends in species richness along alatitudinal gradient and projected trends under climatechange (Currie 2001) for amphibians and reptiles andfor birds and mammals Projected increases anddecreases in richness raise the question of which specieswill expand into newly suitable habitat and whichspecies will go extinct in increasingly unsuitable habitat

Wholoses

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

Much work is needed on methods to execute and evaluatesuch strategies Simulation models and other decision-support tools are needed to assess the likely response of a com-munity to global change and to evaluate the potential successof alternative management strategies It is especially impor-tant that these models integrate consideration of climateland use and biodiversity Adequate monitoring protocols areneeded to establish rates of change in environmental driversand species and community responses to these changesAdaptive management experiments can be used to evaluatethe success of management manipulations These experi-ments should include manipulations of species distributionsand performance through planting and release projects habi-tat modification genetic engineering and eradication of un-desirable species In addition species banks or refuges for col-onization can be developed

Decisions on mitigation of the effects of climate change onbiodiversity involve social political and economic consider-ations Therefore these decisions will not be made solelywithin the ecological context of the issues Instead the full setof advantages and disadvantages of all decisions must beconsidered However choosing not to control greenhousegases or not to manage species communities and landscapesin the face of climate change is making a decision about theimpact of these changes on biodiversity Rather than lettinginaction decide the result of potential changes active assess-ment and management will be necessary to bring the land-scape to a state of desired future biodiversity conditions in theface of global climate change

Research needsAs is apparent from the discussion above aspects of globalchange research have substantial levels of uncertainty Al-though considerable progress has been made toward under-standing global change major research initiatives will be re-quired to reduce current uncertainty

Future climate and land use The predictions of cur-rent GCMs disagree to varying degrees and their levels of ac-curacy are not well quantified The models could be better val-idated relative to past conditions at local to regional scales Thiswould allow research and management to focus on the futureclimate scenarios that are most plausible In contrast to cli-mate relatively little effort has gone into understanding andpredicting land-use and land-cover change Studies are neededto project land cover and use based on biophysical factors andsocioeconomic factors Moreover integrated models of climateand land use are needed to better predict future interactionsbetween these two aspects of global change

Tolerances of organisms Much is known about the tol-erances of some trees to climate soils and other biophysicalcontrols However such knowledge is mostly lacking forother plants vertebrates and invertebrates Similarly little isknown about the demographic performance of organismsacross gradients in climate and land use (Hansen and Rotella

forthcoming) This deficiency is especially apparent for pop-ulation dispersal Species at higher trophic levels will proba-bly be more difficult to model under global change becausethey respond directly to climate as well as indirectly to the sec-ondary effects mediated by habitat structure ecological pro-ductivity and interactions with other species Thus holisticexaminations of species and environmental relationships areneeded that consider multiple stressors and multiple spatialand temporal scales

Biodiversity feedbacksWe emphasize that organisms me-diate the effects of global change on ecosystems and feedbacksto climate and land use What might be the consequences onthe services provided by ecosystems of the changes in biodi-versity predicted under global change Knowledge of the roleof biodiversity in ecosystem function is underdevelopedMore research is needed on how species composition feedsback to influence climate and land use

Mitigation strategies We have presented some of thetypes of management strategies that will be needed to copewith global change Actual development and evaluation of al-ternative techniques for moving species managing distur-bance controlling exotics and other coping strategies meritconsiderable attention

ConclusionsAll ecological systems are dynamic and variations in climatedisturbance and other ecological processes are required formaintaining some species and communities However changein biodiversity over the last century has been accelerated sub-stantially by human land use and possibly by human-inducedchanges in climate Rates of change in biodiversity are likelyto be even greater in the near future Our generation and thenext one face the novel situation of having prior knowledgeof the impending change We also have increasingly sophis-ticated sets of data and tools for understanding and manag-ing this change How we use this knowledge and these re-sources may determine our well-being under global changeWe draw the following conclusions from this review

middot Land use and to a lesser extent climate have changedsubstantially over the past century causing importantshifts in the abundance and distribution of speciescommunities and biomes

middot Future changes in climate and land use are likely to beof a magnitude that cause even greater changes in bio-diversity The distributions of some species communi-ties and biomes are likely to expand while others con-tract and entirely new communities of species mayform

middot The resulting changes in biodiversity are likely to feedback and influence land use climate and human well-being

September 2001 Vol 51 No 9 BioScience 777

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

middot There is considerable uncertainty in the magnitude and

in some cases the direction of climate and land-use

change expected in the future as well as in the respons-

es of ecosystems and organisms The pace of land-use

and climate change is likely to be rapid relative to the

adaptability of species leading to rapid shifts in species

ranges extinctions and disequilibrum ecosystem

dynamics ldquoConsequently the only outcome that can be

predicted with virtual certainty is major surprises The

only forecast that seems certain is that the more rapidly

the climate changes the higher the probability of sub-

stantial disruption and surprise within natural systemsrdquo

(Root and Schneider 1993 p 267) Substantial invest-

ment in research and assessment will be needed to

reduce uncertainties in the interactions between cli-

mate land use and biodiversity

middot Some level of uncertainty will remain however and

policymakers and managers will benefit from incorpo-

rating consideration of this uncertainty in future out-

comesmdashand the risk of those outcomesmdashinto their

decisionmaking

middot Current thinking on strategies and methods for coping

with global change is underdeveloped A comprehensive

program of research planning and adaptive manage-

ment would better allow society to understand manage

and cope with global change before the changes erode

biodiversity and human well-being

middot Because of the spatial scale of global change coping

strategies will require a new level of cooperation among

public and private land stewards and among nations

AcknowledgmentsThis paper was prepared by the Biodiversity Committee of theForest Sector of the National Assessment of the PotentialConsequences of Climate Variability and Change We thankForest Sector Chairs John Aber and Steve McNulty and theForest Sector team for informative discussions Helpful com-ments on the manuscript were provided by John Aber LisaGraumlich John McCarty and two anonymous reviewersFunding was provided by the US Global Change Research Pro-gramrsquos National Assessment of Climate Change the USDAForest Service Global Change Research Program the NASALand Cover Land Use Change Program and the Depart-ment of Energy The national assessment was mandated bythe US Congress to provide detailed understanding of the con-sequences of climate change for the nation and to examine thepossible coping mechanisms to adapt to climate change OakRidge National Laboratory is managed by the University ofTennesseendashBattelle LLC for the US Department of Energy un-der contract DE-AC05-00OR22725

References citedAber J Neilson RP McNulty S Lenihan JM Bachelet D Drapek RJ Forest

processes and global environmental change Predicting the effects ofindividual and multiple stressors BioScience 51 735ndash751

Abraham KF Jefferies RL 1997 High goose populations Causes impacts andimplications Pages 7ndash72 in Batt BJD ed Arctic Ecosystems in Peril Re-port of the Arctic Goose Habitat Working Group Arctic Goose Joint Ven-ture Special Publication of the US Fish and Wildlife Service Washing-ton DC and the Canadian Wildlife Service Ottawa Ontario

Arno SF Gruell GE 1983 Fire history at the forest-grassland ecotone in south-western Montana Journal of Range Management 39 332ndash336

Bachelet D Neilson RP Lenihan JM Drapek RJ 2001 Climate change effectson vegetation distribution and carbon budgets in the US Ecosystems 4164ndash185

Bartlein PJ Whitlock C Shafer SL 1997 Future climate in the YellowstoneNational Park region and its potential impact on vegetation Conserva-tion Biology 11 782ndash792

Beebee TJC 1995 Amphibian breeding and climate Nature 374 219ndash220Beier P 1993 Determining minimum habitat areas and habitat corridors for

cougars Conservation Biology 7 94ndash108Boer GJ Flato GM Ramsden D 2000 A transient climate change simulation

with historical and projected greenhouse gas and aerosol forcing Pro-jected climate for the 21st century Climate Dynamics 16 427ndash451

Brown DE Davis R 1995 One hundred years of vicissitude Terrestrial birdand mammal distribution changes in the American Southwest 1890ndash1990Pages 231ndash244 in DeBano LF Ffolliott PE Ortega-Rubio AGottfried GJHamre RH Edminster CB technical coordinators Biodiversity andManagement of the Madrean Archipelago The Sky Islands of South-western United States and Northern Mexico Fort Collins (CO) USDAForest Service General Technical Report RM-GTR 264

Chase TN Pielke RA Sr Kittel TGF Baron JS Stohlgren TJ 1999 Potentialimpacts on Rocky Mountain weather and climate due to land use changesin the adjacent Great Plains Journal of Geophysical Research 10416673ndash16690

Crick HQP Dudley C Glue DE Thomson DL 1997 UK birds are laying eggsearlier Nature 388 526

Currie DJ 1991 Energy and large scale patterns of animal and plant speciesrichness American Naturalist 137 27ndash39

mdashmdashmdash 2001 Projected effects of climate change on patterns of vertebrateand tree species richness in the coterminous United States Ecosystems4 216ndash225

Dale VH 1997 The relationship between land-use change and climatechange Ecological Applications 7 753ndash769

Dale VH et al 2001 Climate change and forest disturbances BioScience 51723ndash734

Davis MB 1989 Insights from paleoecology on global change Ecological So-ciety of America Bulletin 70 222ndash228

Davis MB Zabinski C 1992 Changes in geographical range resulting fromgreenhouse warming Effects on biodiversity in forests Pages 297ndash308 inPeters RL Lovejoy TE eds Global Warming and Biological Diversity NewHaven (CT) Yale University Press

Dickenson RE 1991 Global change and terrestrial hydrology A reviewTellus 43AB 176ndash181

Drake JA Mooney HA di Castri F Groves RH Kruger FJ Rejmanek MWilliamson M eds 1989 Biological Invasions A Global PerspectiveChichester (UK) John Wiley and Sons

Flather CH Knowles MS Kendall IA 1998 Threatened and endangeredspecies geography BioScience 48 365ndash376

Flather CH Brady SJ Knowles MS 1999 Wildlife Resource Trends in theUnited States A Technical Document Supporting the 2000 USDA For-est Service RPA Assessment Fort Collins (CO) USDA Forest ServiceRocky Mountain Research Station General Technical Report RMRS-GTR-33

Frank DA McNaughton SJ 1991 Stability increases with diversity in plantcommunities Empirical evidence from the 1988 Yellowstone droughtOikos 62 360ndash362

778 BioScience September 2001 Vol 51 No 9

Articles

Franklin JF 1993 Preserving biodiversity Species ecosystems or land-scapes Ecological Applications 3 202ndash205

Franklin JF Forman RTT 1987 Creating landscape patterns by forest cut-ting Ecological consequences and principles Landscape Ecology 15ndash18

Franklin JF et al 1992 Effects of global climate change on forests in north-western North America Pages 244ndash257 in Peters RL Lovejoy TE edsGobal warming and biological diversity New Haven (CT) Yale Univer-sity Press

Gilpin ME 1987 Spatial structure and population vulnerability Pages125ndash140 in MSouleacute edViable Populations for Conservation Cambridge(UK) Cambridge University Press

Halpin PN 1997 Global climate change and natural-area protection Man-agement responses and research directions Ecological Applications 7828ndash843

Hansen AJ Rotella JJ 1999 Environmental gradients and biodiversity Pages161ndash209 in M Hunter ed Managing Forests for Biodiversity Cam-bridge (UK) Cambridge University Press

mdashmdashmdash Biophysical factors land use and species viability in and around na-ture reserves Conservation Biology Forthcoming

Hansen J Fung I Lacis A Rind D Lebedeff S Ruedy R 1988 Global climatechanges as forecast by Goddard Institute for Space Studies three-di-mensional model Journal of Geophysical Research 93 9341ndash9364

Houghton RA 1994 The worldwide extent of land-use change BioScience44 305ndash313

Iverson LR Prasad AM 1998 Predicting potential future abundance of 80tree species following climate change in the eastern United States Eco-logical Monographs 68 465ndash485

mdashmdashmdash 2001 Potential changes in tree species richness and forest commu-nity types following climate change Ecosystems 4 186ndash199

Iverson LR Prasad AM Hale BJ Sutherland EK 1999a An atlas of currentand potential future distributions of common trees of the eastern UnitedStates Delaware (OH) USDA Forest Service Northeastern ResearchStation General Technical Report NE-265

Iverson LR Prasad AM Schwartz MW 1999b Modeling potential future in-dividual tree-species distributions in the Eastern United States under aclimate change scenario A case study with Pinus virginiana EcologicalModelling 115 77ndash93

Johns TC Carnell RE Crossley JF Gregory JM Mitchell JFB Senior CA TettSFB Wood RA 1997 The second Hadley Centre coupled ocean-at-mosphere GCM Model description spinup and validation ClimateDynamics 13 103ndash134

Maizel M et al 1999 Historical interrelationships between population set-tlement and farmland in the conterminous United States 1790 to 1992Pages 5ndash12 in Sisk T ed Land Use History of North America Providinga Context for Understanding Environmental Change Washington (DC)US Geological Survey Biological Resources Division

Malanson GP Cairns DM 1997 Effects of dispersal population delays andforest fragmentation on tree migration rates Plant Ecology 131 67ndash80

Manabe S Wetherald RT Mitchell JFB Melesko V Tokioka T 1990 Equi-librium climate change and its implications for the future Pages 131ndash172in Houghton JT Jenkins GJ Ephraums JJ eds Climate Change The IPCCScientific Assessment New York Cambridge University Press

Mueller-Dombois D 1992 Potential effects of the increase in carbon diox-ide and climate change on the dynamics of vegetationWaterAir and SoilPollution 64 61ndash79

Neilson RP 1995 A model for predicting continental-scale vegetation dis-tribution and water balance Ecological Applications 5 362ndash385

Noss RF LaRoe ET Scott JM 1994 Endangered Ecosystems of the UnitedStates A Preliminary Assessment of Loss and Degradation Washington(DC) US Fish and Wildlife Service

Peters RL 1992 Conservation of biological diversity in the face of climatechange Pages 15ndash30 in Peters RL Lovejoy TE eds Global warming andbiological diversity New Haven (CT) Yale University Press

Pimentel D Wilson C McCullum C Huang R Dwen P Flack J Tran Q Salt-man T Cliff B 1997 Economic and environmental benefits of biodiversityBioScience 47 747ndash757

Pimm SL 1991 The Balance of Nature Chicago University of Chicago PressPrasad AM Iverson LR 1999 A Climate Change Atlas for 80 Forest Tree

Species of the Eastern United States (6 August 2001wwwfsfedusnedelawareatlas)

Prentice KC 1990 Bioclimatic distribution of vegetation for general circu-lation model studies Journal of Geophysical Research 95 11811ndash11830

Primack RB Miao SL 1992 Dispersal can limit local plant distribution Con-servation Biology 6 513ndash519

Pulliam HR 1988 Sources sinks and population regulation AmericanNaturalist 132 652ndash661

Rapport DJ Regier HA Hutchinson TC 1985 Ecosystem behavior understress American Naturalist 125 617ndash640

Rogers DJ Randolph SE 2000 The global spread of malaria in a futurewarmer world Science 289 1763ndash1766

Rogers R 1990 Quercus alba L White Oak Pages 605ndash613 in Burns RMHonkala BH coordinators Silvics of North AmericaVol 2 HardwoodsWashington (DC) USDA Forest Service Agriculture Handbook 654

Root TL Schneider SH 1993 Can large-scale climatic models be linked withmultiscale ecological studies Conservation Biology 7 256ndash270

Schlesinger ME Zhao ZC 1989 Seasonal climatic change introduced by dou-ble CO2 as simulated by the OSU atmospheric GCMmixed-layer oceanmodel Journal of Climate 2 429ndash495

Shafer SL Bartlein PJ Thompson RS 2001 Potential changes in the distri-butions of western North America tree and shrub taxa under future cli-mate scenarios Ecosystems 4 200ndash215

Swanson FJ Dyrness CT 1975 Impact of clear-cutting and road construc-tion on soil erosion by landslides in the western Cascade Range OregonGeology 3 393ndash396

[USDI] US Department of the Interior Fish and Wildlife Service 2000Listings and recover plans as of June 30 2000 Endangered Species Bul-letin 25 40

VEMAP members 1995 Vegetationecosystem modeling and analysis pro-ject Comparing biogeography and biogeochemistry models in a conti-nental-scale study of terrestrial ecosystem responses to climate changeand CO2 doubling Global Biogeochemical Cycles 9 407ndash437

Vitousek PM Ehrlich PR Ehrlich AH Matson PA 1986 Human appropri-ation of the products of photosynthesis BioScience 36 368ndash373

Wardle DA Huston MA Grime JP Berendse F Garnier E Lauenroth WKSetala H Wilson SD 2000 Biodiversity and ecosystem function An is-sue in ecology Bulletin of the Ecological Society of America 81 235ndash239

Watson RT Zinyowera MC Moss RH eds 1998 The Regional Impacts ofClimate Change An Assessment of Vulnerability IntergovernmentalPanel on Climate Change New York Cambridge University Press

Webb T III 1992 Past changes in vegetation and climate Lessons for the fu-ture Pages 59ndash75 in Peters RL Lovejoy TE eds Global warming and bi-ological diversity New Haven (CT) Yale University Press

Wilcove DS Rothstein D Dubow J Phillips A Losos E 1998 Quantifyingthreats to imperiled species in the United States BioScience 48 607ndash615

Wilson CA Mitchell JFB 1987 A doubled CO2 climate sensitivity experimentwith a global climate model including a simple ocean Journal of Geo-physical Research 92 13315ndash13343

Wright DH Currie DJ Mauer BA 1993 Energy supply and patterns of speciesrichness in local and regional scales Pages 66ndash74 in RE Ricklefs DSchluter eds Species Diversity in Ecological Communities Chicago Uni-versity of Chicago Press

September 2001 Vol 51 No 9 BioScience 779

Articles