Global Warming and Extinctions of Endemic SpeciesJAY R MALCOLMtt CANRA - PDF document

1 Global Warming and Extinctions of Endemi
Global Warming and Extinctions of Endemic SpeciesJAY R. MALCOLM,*tt CANRAN LIU,*t RONALD P. NEILSON,* LARA HANSEN,§*Faculty of Forestry, University of Toronto, Toronto, ON, M5S 3B3, CanadatDepartment of Ecosystem Management, University of New England, Armidale, NSW 2351, AustraliaOR97331, U.S.A. Global warming is a key threat to biodiversity, but few researchers have assessed the magnitude2climates, calculated changes in habitatnumerous uncertainties in this approach, we undertook a sensitivity analysis of multiple factors that included(3)different assumptions about whether species distributions were biome specific or not, and (4) differentdoubled-CO2climatefferent from effects on random same-biome collections of2000species. Under the assumption thatprojected habitat changeswere attained in 100 years, estimated global-warming-induced rates of species extinctions in tropical hotspotsKey Words: biomes, climate change, general circulation models, global vegetation models, migration, speciesextinctionsResumen:El calentamiento global es una amenaza cla

2 ve para la biodiversidad, pero pocos inv
ve para la biodiversidad, pero pocos investigadoreslaproyeccidn de la distribucidn futura de los biomas en2,ciones de especies de plantar y animales endemicas en sitios de importancia para la biodiversidad. Debido a2004;revised manuscript accepted October20, 2005.538Conservation Biology Volume 20, No. 2, 538-548©2006Society for Conservation Biology10.1111/j.1523-1739.2006.00364.x Malcolm et al.Global Warming and Extinction(3)suposiciones diferentes sobre si la distribuci6n de especies era especifica de un2en 100 anos.bioma o de tasas de migraci6n; sin embargo, en algunos escenarlos los sitios de importancia para la biodi-de importancia para la biodiversidad especialmente vulnerables fueron la Region Floristica del Cabo, Caribe,Palabras Clave: biomas, cambio climatico, extinciones de especies, migracion, modelos generates de circu- IntroductionGlobal warming represents perhaps the most pervasive oftheir high species richness and endemism (e.g., Mitter-ber of the world's species (some 44% of the world's plants2concentrationsattempts to assess biodiv

3 ersity impacts. Biomes describeables tha
ersity impacts. Biomes describeables that are relevant to a wide range of organisms andmay be especially valuable in areas where species distri-Conservation Biology 540in this approach, we also undertook sensitivity analyses of grid cells from the Earth's terrestrial surface that hadMethodsBiomes were modeled using GVMs under recent (1 x C02)2) climatic conditions, thevariation among the models. We reasoned that if a broadused a suite of 14 combinations of two GVMs and sevenof variation among the 14 model combinations with re-2climates (Malcolm et al. [2002] used2,Conservation BiologyVolume 20, No. 2, April 2006The GCMs included both older-generation models thatused simple mixed-layer oceans (GISS, GFDL-R30, OSU,2and 2x CO2simulations to a baseline (1x C02)2effects, whereas in keeping with the VEMAP analyses, for2effects (see Neilson et al. [1998] for additional details).Habitat LossA second area of uncertainty was the method used totat loss with loss of primary habitat (following Myers etbreadth of habitat conditions that a species can toler-may

4 migrate to novel locations. Accordingly,
migrate to novel locations. Accordingly, we estimatedbiogeographic barriers (such as oceans and topographic Malcolmetal.Global Warming and Extinction541Table 1. Sensitivity analysis of numbers of extinctions of endemic plant and vertebrate species from 25 biodiversity hotspots under scenarios of GlobalMigrationBiomeBiomeb(9,645 endemics)b EARSAREARSAR PerfectbroadbroadBIOME3narrowBIOME3narrowBIOME3narrowBIOME3ZerobroadBIOME3narrowBIOME3MAPSSBIOME3narrowBIOME35,502 (4)4,512 (3)8,300 (6)4,354 (3)8,790 (7)26,110 (20)56,606 (43)3,829 (40) 'Extinctions are shown for two migration scenarios, two biome breadth definitions, two levels of biome specificity, and two global vegetationbPercent loss of endemic species in parentheses.for a given amount of climate change (and hence lowerSeasonal Forest (which itself was distinguished from theTropical Deciduous Forest).ing biomesasproxiesforthe overall climate envelope2con-ditions and, for those biomes that showed a reduction in2climates, summed the50%Savanna/Woodland butareaof Tropical Broadleaf Forest).a, b,and

5 c in proportionspa, pb,andpe,then the p
c in proportionspa, pb,andpe,then the proportionawas estimated aspa/(pa+b+pcz).In the above example, Savanna/Woodland and2climate). Thus, in calculating projected habitatConservation Biology 542Global Warming and Extinctionthe two migration scenarios we calculated required mi-rates by dividing by 100 years, which is a conservative2clihotspot.2conditions had theSpecies ExtinctionsIn addition to the standard species-area relationship wealso calculated species loss using the more conservativeships in some cases overestimate the number of extinc-exceeds 50%, endemic-area estimates and species-area es-Following Kinzig and Harte (2000), under species-areatoof the habitat Ao ofahotspot will lead to the extinction of a fraction f of the(lost-SAR=1-1-`l1I'A)z,where z is the species-area exponent and SAR is the2~fiost-EAR = (A ,where'= - ln(1 - 1/2z)/ 1'1(2)-Conservation Biologyhence instead of a z value typical of fragmented habitatsResultsProjected extinctions varied markedly depending on as-(p= 0.0001; 3.9-fold change in means),19% (p= 0.007;As expected

6 , endemic-area-based estimates were smal
, endemic-area-based estimates were smallarea-based estimates converged. Averaged across all fourspecificity only). Two hotspots that showed consistentlyChile and, to a lesser extent, Wallacea. Other hotspotsthat tended to show low amounts of change relative toNew Zealand. In contrast, Tropical Andes, Southwest Aus- Malcolm et al,andExtinctionTable 2. Percent habitat loss from 25 biodiversity hotspots under various scenariosofglobal warming (two migration scenarios, two biome breadthdefinitions, and two global vegetation models [BIOME3 and MAPSS]) assuming broad biome specificity.Perfect migration')HotspotNumber ofdefinitiondefinitionMAPSSBIOME3Atlantic Forest26242232(18)(16)(42)(30)(64)811924(14)(24)43(29)**17)*(40)***4042(10)3717213936(23)362254()(0)(15)(18)(34)3(18)64210(49)46Sundaland(4)50**52***63***29***(13)(9)(18)(18)(25)(47) "Grid-cell dimensions were 0.5€ latitude by 0.5€ longitude."Parentheses indicate whether or not mean biome change was greater than (no parentheses) or less than (parentheses) bootstrap-based globalexpectations. Mean si

7 gnificance levels comparing the observed
gnificance levels comparing the observed and bootstrap means: *p **p**pthan expected. Other hotspots that tended to show rel-&#x 0.0;
.0;p 0.05, two-tailed binomial test). The exception was the sce-gration rates (100-200 in/year) to rates well in excess oflowmigration rates outnumbered thosewith high migration rates, although significantly so onlyConservation BiologyNo. 2, April 2006 544Global Warming and ExtinctionMalcolm et al,Table 3. Mean required migration rates (meters per year) in 25 biodiversity hotspots under scenarios of global warming based on two biomeBroad biome definition","MAPSSBIOME3MAPSS309(315)(455)California Floristic Province(175)*(267)(667)Cape Floristic Region(206)1,004(2,224)Caribbean2631,452*(545)2,117*Caucasus(254)***Central Chile(193)*(265)*(414)(1,078)Brazilian Cerrado(231)Choco-Darien-Western Ecuador174(273)Eastern Arc Mountains & Coastal Forest(228)(408)(542)1,516Indo-Burma2304631,228(681)Madagascar & Indian Ocean Island(144)(264)(952)Mediterranean Basin(294)Mesoamerica257New Caledonia(0)11,119New Zealand(188)1,460

8 (188)2,562Philippines(59)Polynesia & Mic
(188)2,562Philippines(59)Polynesia & Micronesia(104)12,95915,660***Mountains of South Central China(339)***(484)******(929)Succulent Karoo(242)(317)(499)(1,036)Sundaland(57)216Southwest Australia1,915***1,157**2,399***2,731Tropical Andes(276)***(514)(690)*Guinean Forests of West Africa(68)Wallacea(60)Western Ghats & Sri Lanka(118)(348)* "Parentheses indicate whether or not mean biome change was greater than (noparentheses)or less than (parentheses) bootstrap-based global*p**p***pMountains of South Central China, and Tropical Andes;2climates were 39-43% of the biota, repre-Conservation BiologyBurma, and Tropical Andes). Our bootstrap comparisonslowre-climate to be individualistic. Where wholesale changes in vege-Discussion Malcolm et at.Table 4. Projected species-area-based percent extinctions of endemic species in 12 hotspots judged to be especially vulnerable to global warming.'Perfect migrationbroad biomenarrow biomebroad biomenarrow biomebvegetationspecificityCalifornia Floristic3.04.07.853.52.34.55.82.72.852.443.921.968.0Caribbean3.64.43.87.24

9 8.5BIOME318.82.740.56.4(13,000; 235)3.74
8.5BIOME318.82.740.56.4(13,000; 235)3.74.45.69.944.30.00.00.00.00.00.075.05.32.55.32.84.63.88.23.543.57.010.1Southwest Australia9.85.310.117.238.718.128.2 "Percentages are shown for two migration scenarios, two biome breadth definitions, two levels of biome specificity, and two global vegetation(BIOME3and MAPSS).bNumbers of endemic plant and vertebrate species, respectively, are shown in parentheses below hotspot names. with rates of habitat loss due to deforestation, which isthe planet's biodiversity. Brooks et al. (2002) estimate an-nual rates of deforestation for 13 tropical forest hotspots.2cli-2climate could oc-similar to that based on deforestation. These calculationsthe planet's biodiversity and, under some scenarios, maying presumably ranks as the most serious threat to biodi-taxa atthe highConservationBiology2,April2006 546Global Warming and Extinctionwhether or not species distributions were specific to cer-ininfluencing extinctions (seealso Peterson et al. 2002), observing a 2.0-fold increasetion capabilities remain poorly understood (P

10 itekia et al.1997; Clark 1998; Malcolm e
itekia et al.1997; Clark 1998; Malcolm et al. 2002). The rapid ratesof migration considered here will presumably be exacer-ing extinction rates was expected, more surprising wasthe GVMs reflect both differences of opinion regardinginusing a biome approach to estimate habitatand the climatic drivers of biome change are similarlyiportant in driving species responses. Under this last as-formation on species distributions (and responses) to beConservation Biologyto vary both the amount of detail in biome definitions 2climates)than their observed ranges would indicate. Different as-Loehle & LeBlanc 1996; Davis 1998a, 1998b; Pearson &those modeled using the climate-envelope approach.Our estimates of global-warming-induced extinctionsmate change poses a serious threat to global biodiversity. Malcolm et at. of thousands, of hotspot endemic plant and vertebrateability to climate change. Although some hotspots appearreduced likelihoods that species will be able to migrate change. Empirical studies that test climate-envelope ap-examine biome-level habitat as

11 sociationsAcknowledgmentsWe thank R. Gre
sociationsAcknowledgmentsWe thank R. Green, B. Huntley, A. Markham, G. Midgeley,of the reviewers. Funding was from World Wildlife Fund,Achard, F., H. D. Eva, H. Stibig, P. Mayaux, J. Gallego, T. Richards, andsessile organisms: a simulation model with measurable parameters.Journal of Vegetation Science 7:831-846.80:417-427.Haxeltine, A., and I. C. Prentice. 1996. BIOME3: an equilibrium ter-source availability, and competition among plant functional types. Contributions of Working Group Ito the third assessment report ofsity Press, Cambridge, United Kingdom.Biogeography29:835-849.United Kingdom.climate change. Conservation Biology 15:578-590.Volume 20, No. 2, April 2006 548Global Warming and ExtinctionPitelka, L. F., et al. 1997. Plant migration and climate change. AmericanScientist85:464-473.Root, T. L., J. T. Price, K. R. Hall, S. H. Schneider, C. Rosenzweig, andConservation 2:51-61.VEMAP (Vegetation/Ecosystem Modeling and Analysis Project) Mem-bers. 1995. Vegetation/ecosystem modeling and analysis project:2doubling. Global Biogeochemical Cycles 9