Individualistic Species Responses Invalidate Simple Physiological Models of Community Dynamics under Global Environmental Change Author(s): Andrew J. Davis, John H. Lawton, Bryan Shorrocks and Linda S. Jenkinson Reviewed work(s): Source: Journal of Animal Ecology, Vol. 67, No. 4 (Jul., 1998), pp. 600-612 Published by: British Ecological Society Stable URL: http://www.jstor.org/stable/2647281 . Accessed: 21/08/2012 01:02 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

British Ecological Society is collaborating with JSTOR to digitize, preserve and extend access to Journal of Animal Ecology.

http://www.jstor.org

JournalofAnimal Ecology)1998, 67, 600-612

Individualistic speciesresponsesinvalidatesimple physiological modelsofcommunity dynamicsunder globalenvironmental change ANDREW LINDA

J. DAVIS,

JOHN H. LAWTONt,

BRYAN SHORROCKS

and

S. JENKINSON

Ecology& Evolution,School ofBiology,The University, Leeds, WestYorkshir-e LS2 9JT, UK, and tNERC Cen1tre for Popuilation Imnper iaclCollegeSiliwoodPark,Ascot,Ber-kshiire SL5 7P Y, UK Biologjy,

Summary 1. Most predictionsof species distributionand abundance changesin responseto global warmingrelate the individualrequirementsof a singleisolated species to climatevariablesthroughsomeformofclimatemapping.Thismethodfailsto account fortheeffects ofspeciesdispersaland speciesinteractions, bothofwhichmaystrongly affectdistribution and abundance. 2. We thereforeexaminedthe effectsof dispersaland species interactionson the distribution and abundance of threeDrosophila speciesin a laboratorysystemthat how speciesdistribution mimickeda latitudinalclineof 15CC. We theninvestigated and abundancein thissystemrespondedto simulatedglobal warming. 3. Dispersalallowedpopulationsto persistat non-optimum temperatures, overriding physiologically imposedrangelimits. 4. Temperaturedeterminedthe outcome of competition.In pairwiseinteractions, Drosophila subobscuiraeliminated D. inelanogasteror D. sitnulansat low temperatures butwas itselfeliminatedat hightemperatures. 5. Competitiveinteractionschanged abundance and range sizes thus shiftingthe positionof species optima. These changes dependedon both the numberand the identityof thecompetingspecies. 6. Enemy-victim interactionsalteredrange and abundance. Adding the parasitoid thehostassemblagedirectly at hightemperatures where Leptopilina boulardi affected theparasitoidwas present,and indirectly (mediatedbydispersal)at low temperatures whereit was scarceor absent.Host speciescoexistedforlongerat low temperatures in clineswhenparasitoidswerepresentthanwhentheywereabsent. 7. Simulatedglobal warmingproduced complex,counter-intuitive effectson distribution and abundance,includingthereversalofspecies'relativeabundanceat some temperatures.

8. Because dispersaland species interactionsstronglyinfluencedboth range and abundance(sometimesin unexpectedways)currentspeciesdistributions are no guide to what theymightbe underglobal climatechange.Furthermore, since both these factorsare missingfromclimateenvelopemodels of rangeand abundance change, theirpredictionsare,at best,incomplete. Key-wovrds: abundance,climatechange,cline,range,temperature. Journalof AniMiialEcologJy(1998) 67, 600-612

Introduction (C 1998 British Ecological Society 600

Almostall attempts to predicttheeffect of global on species'rangesandabundance makeuse warming of'climate & Maywald1985;Nix mapping' (Sutherst

1986; Kohlmann,Nix & Shaw 1988; Rogers& Ran& Jeifree1996).This dolph 1993;Porter1995;Jeifree is thevenerable(Latreille1819)and attractively simple idea that range and abundance are determinedby the individualorganism'sphysiologicalresponseto

climate,particularlytemperature.It remainsinfluential(Hoffmann& Blows 1994) and has beenwidely used in ecoclimatics(Rogers 1979; Caughley etal. 1987; Parry 1990; Sutherst1990; Carter, Parry & Porter 1991; Porter,Parry& Carter 1991; Scott & Poynter1991; Beerling1993; Jeffree & Jeffree 1994). If it were true,species rangeswould simplyfollow shiftsin theircharacteristicclimate envelopes and changes due to global warmingcould be predicted easily,givenadequatemodelsoffutureclimate.Populationsof organismsare,however,affectedby a multitudeofadditionalfactorswhichcomplicatethissimple view(Lawton 1995),especiallythosethatproduce horizontaland verticalintegrationof populations. Predictionsfromclimatemappingignorethe potentiallymost potentof these:speciesdispersaland the multitudeof complexspeciesinteractions, themselves affectedby climate (Kingsolver 1989) and already known to influencerange and abundance (Connell 1961;Paine 1966;Lawton& Hassell 1984;Price1992). The importanceof dispersalis thatit disruptsthe dependenceof populationson locallyactingclimate, allowinglocal presenceand abundance to be influenced by other,distantpopulations.This, in the real world, creates networksof interdependentpopulationsthatinfluenceeach other.In metapopulations (Gilpin & Hanski 1991; Hanski & Gilpin 1997) and central-marginalpopulations (Boorman & Levitt 1973;Brussard1984)withsource-sink dynamics(Pulliam 1988),local presenceand abundanceare strongly influencedby dispersalbetweenpopulations.Many sinkpopulationsmay onlypersistiftheyreceivesustained immigrationfrom sources elsewherein the range(Watkinson1985; Rodriguez,Jordano& Fernandez Haegar 1994). Under such conditions,widespreadin nature,dispersalwillalterrangesand abundance predictedfromthe physiologicalresponsesof individualspecies. Similarly,both horizontal(competition)and vertical (enemy-victim)interactionswill affectspecies because these interactions ranges and distributions, are unlikelyto remainunchangedunderglobalwarmaffectedby climate ing. If species are differentially change, as climate mapping sometimes predicts & Jeffree (Jeffree 1996),theirinteractions maybe significantlyaltered or even totallydisrupted(Peters 1992; Walter & Patterson1994) leading to the dissociationof currentspeciesassemblagesand the formationof newones. That suchfracture has beenproduced by individual species' responses to climate change in the past is supportedby an increasing amountof evidenceforboth animals(Graham 1992; Valentine& Jablonski1993; Coope 1995) and plants (Davis 1981, 1986; Overpeck,Webb & Webb 1992; Brubaker& McLachlan 1996). It is likely,therefore, that climatechange will produce, not replicatesof (C 1998 British currentassemblages,butnewconstellations ofspecies Ecological Society (Walter & Patterson 1994), 'non-analogue' associJournalofAninmal Ecology,67, 600-612 ationsabsentfromthecurrentbiota (Roy etal. 1996).

601 A.J. Davis et al.

will It is unlikelythattheeffects of theseinteractions be confinedto theedge of speciesranges,as has been suggested(Sutherst,Maywald & Skarratt1995),and realisticassessmentsofrangeand abundancechanges requireconsiderationof the overallpatternof interrelationships (Cammell& Knight1992). by Because distribution and abundanceare affected predictionsbased dispersaland speciesinteractions, on climatemappingmustbe seen,at best,as simplya 'null model' of expectedspecies displacementwith climatechange.In practice,this'null model' is likely to be in errorto an unknowndegreebecauseitignores of distribution thebioticeffects and abundance. In thestudyreportedhere,a simplelaboratorysystemwas used to demonstrate a possiblegeneralparadigmforthe responseof speciesassemblagesto climate change. Dispersal, competitionand enemyvictim interactionsare shown to have important effects on distribution and abundanceand, in systems theeffects ofglobal wherethesefactorsare important, notto be thesimpleconsewarmingare demonstrated responses. quences of individualspeciestemperature Species rangesand abundance in systemsincluding along experimental dispersaland speciesinteractions temperatureclines are shown to be substantially fromthose suggestedby the physiologyof different single species and, consequently,range and abundance in assemblagescannotbe predictedfromthose recordedforsinglespecies.The studydemonstrated thattheresponsesofsimpleassemblagesto simulated globalwarmingwerenotpredictablefromthespecies' distributionand abundance on the clines before warming.

Methods EXPERIMENTAL

ORGANISMS

Three Drosophilaspecieswereused: D. nielanogaster Meigen, D. sim1ulans Sturtevantand D. stbobscur7a Collin, and a parasitoid wasp Leptopiliniaboula7di (Barbotin,Carton & Kelner-Pillaut).The responses ofthisgroupingare likelyto be paradigmaticofmany animal assemblages,as the threeDrosophilaspecies do not have specializedlifehistoriesor larval diets. the four species are an elementof a Furthermore, naturallyoccurringspecies assemblage. Drosophila occur fromthetropics and D. simnulans nielcanogaster to, in Europe,about latitude54?N. DrosophilasuibobscuraoccursfromNorthAfricato theArcticbutdoes notoccurinthetropics(Wheeler1981).The parasitoid is warm-adapted,occurringin the tropicsand subas faras latitude tropics(Nordlander1980),extending 48?N in Europe(Cartonetal. 1991)butabsentfurther north(Davis etal. 1996). It therefore co-occurswith the threeDrosophilaspeciesin southernand central Europe and theMediterraneanbasin. The parasitoid

preferentially oviposits in D.

m1elan1ogaster, in which

itsuffers littlemortality, butwillalso attackand cause

602 Individualistic responsesto climatechange

Table 1. Summaryof thetermsused forthedifferent experimental systems Term

System

Series Cline

Set of eightcages connectedtogetheracrossfourincubatorsat thesame temperature Set ofeightcages connectedtogether(two/cage)acrossfourincubatorsat fourdifferent consecutivetemperatures Cline withincubatorsat 100, 150, 200 and 25?C Cline withincubatorsat 150, 200, 250 and 30?C Cline in whichthe tubesconnectingcages betweenincubatorsare blockedwithfoam bungs Cline in whichthetubesare not blockedand fliescan disperse D. simnulans or D. Seriesor clineinitiatedwitha singlespecieseitherD. nielaniogaster, a suibobscur Clineinitiatedwiththetwo speciesD. simulans and D. subobscura ClineinitiatedwiththethreespeciesD. inelanogaster, D. sirnulansand D. subobscura ClineinitiatedwiththethreeDrosophilaspeciesand theparasitoidLeptopilinabotulacrdi

Cold cline Hot cline Closed cline Open cline Singlespeciesseries(or cline) Two speciescline Threespeciescline Four speciescline

highhostmortality in D. siniulansand D. subobscur a, even thoughthe wasp larva rarelysurvivesin these species(Cartonetal. 1986;Kraaijeveld& van Alphen 1995). The Drosophilaspecieswereall collectedfromPontefractLane fruitmarket,Leeds, UK, and had been maintainedin laboratorymass culturefor 2 years beforebeingused intheexperiments. The samestrains The parasitoidwas wereused in all the experiments. collectedin Tasagil,Turkey,and was mainoriginally tainedthroughouton D. mnelainogaster. EXPERIMENTAL

ECOSYSTEMS

Experimental systemsconsistedofeightPerspexcages (each 100 x 150 x 300mm) linked in series (see Table 1 forterminology). The eightcages were connectedtogetherin pairs and each pair housed in a incubator.The tubingconnectingthe cages different was 30 mm in diameterbecause this size restricted dispersal.Up to fivereplicatesystemswere accommodatedin each groupoffourincubators(Fig. 1) and thereweretwosuchgroups.The incubatorsin a group were initiallyall set to a nominal 20?C and temperatureclineslatercreatedby settingtheincubators

sequentiallyto 10, 15, 20 and 25?C ('cold' clines) representing thecurrentspreadofmeansummertemperaturesfromsouthernSpain to northernEngland. To simulateglobal warmingthe overalltemperature of theclineswas raisedby 5 ?C (Bennetts1995) to 15, 20, 25 and 30 ?C ('hot' clines).Temperatures measured byprobesinsidethecages did notdeviatesignificantly fromthe incubatorsetting(mean deviation= 0 28 SD

=

0445 t = 0626, d.f. = 77, P>

005). The over-

all gross variabilityof thesetemperatures, including incubatorfaultsand veryoccasional power failures, was about 2?C (95% confidenceinterval(CI) = 1 78?C) withmost of the variabilityconcentratedin the lowest temperatureincubators (95% CI = 4 50?C). In normalrunning,however,temperatures wereless variable(95% CI = 0 88?C). Each of the cages containedsix food tubes which were replaced sequentiallywith a freshtube containing50 mL standard,cereal-basedDrosophilamedium (Shorrocks1972). The cages werereprovisioned at the rate of one tube everyhalf generation,thus, because generationtimeis inverselyrelatedto temperature,hottercages werereprovisioned moreoften but Dr-osophilaat the four different temperatures receivedthesame amountof food per generation.In

cl

? 1998British Ecological-Society Journ7lal ofAnimilal Ecology,67, 600-612

12 11 13 14 Fig. 1. The arrangement of Perspexcages (100 x 150 x 300mm) in one set of fourincubators(I1-14) showing,shaded,an seriesof eightcages linkedtogetherby tubing.Each pair of cages (e.g. cl-c2) experiencesthesame temperature. experimental

603 A.J.Davis et al.

four-speciesclines honey was providedas food for adultL. boulardi. POPULATION

MEASUREMENT

Experimentalpopulationswere assessed by weekly counts of the adult Dr-osophilaon a standard 80 x 80 mm gridfixedon the rearwall of each cage. Because D. inelanogasterand D. siniullans are morphologicallyverysimilar(e.g. Patterson1943; Moore 1952) and cannot be distinguishedwithoutdetailed examination,the standardcounts were adjusted by theproportionsofthesetwospeciespresentin weekly samplesof 100-150fliesdrawnfromeach cage. Estimatesof thereal populationsin each cage werethen made by applyinga seriesof calibrationcurvesto the countdata. These curveswereproducedby releasing knownnumbersofflies(between50 and 1000foreach speciesand each temperature) intoexperimental cages and countingthenumberson thestandardgrid.This partialstandardcountwas closelycorrelatedwiththe real numberof adult fliesbut the linearrelationship overestimated numbersat boththelowestand highest densities.Biologicallymore realisticsigmoidcurves [withgeneralformy = k/(l + e"-()], whichreflected thespatialbehaviourof thefliesweretherefore fitted separatelyto the counts for each species and temperature(Table 2). In two of the 15 calibrations,for D. simltlansand D. subobscuraat 30?C, the sigmoid curveprovidedno betterfitbetweenrealnumbersand counts than did the overall mean numberof adult flies.Consequently,experimentalpopulationsof D. simnulansat 30?C wereestimatedas the overallmean forD. subobsctura, butthiswas unnecessary whichdid not occur at 30 'C. To avoid the initialphases of populationincreaseand to standardizethetimeperiod consideredforeach treatmentthe analyseshere are confinedto countsmade betweenweeks5 and 25. CAGE

SERIES

AND

CLINES

The distributionand abundance of each species of Drosophilawas firstassessedwhendispersalwas possible betweenincubatorsbut in the absence of tembetweenincubatorsand without peraturedifferences interspecificinteractions.Nine one-species series,

three replicatesfor each species, were initiatedby adding25 male and 25 femalefliesto all eightcages in a series(a totalof 400 fliesperseries).The start-up dates of these replicates,and for those in all other treatments,were staggeredto avoid alterationsin externalconditionscausingcorrelatedchangesacross replicates.To determinethe effectof a temperature the discline on singlespecieswithoutinteractions, tributionand abundanceof the threespeciesin onespecies cold clines were then remeasured.The cage serieswereconvertedto clinesby resetting the incubatorsto thespecifiedcold-clinetemperatures. DISPERSAL

A featureof this studyis the explicitinclusionof dispersalin theexperimental system.However,itproved impossibleto assess dispersalratesin the clines accurately by directly monitoring flies moving between temperaturesbecause numerousflies circulatedin the connectingtubeswithoutenteringthe carriedout cages at eitherend. Tests weretherefore to discoverwhetherdispersalhad a major effecton the distribution and abundanceof individualspecies in the systemby preventingdispersalaltogetherin one-speciesclines.Nine closed clines,threereplicates foreach species,werecreatedby blockingthe tubes betweenincubatorsin establishedcold one-species open clines. INTERACTIONS

The effectsof two kindsof interspecific interactions wereexamined;thosebetweenDrosophilaspeciesand those involvingenemy-victim interactionsbetween and theparasitoidL. bo-tlar-di. Dr-osophila Interactions between Drosophila species were assessedin twoways.First,to determine theoutcome of interactionsbetweenpairs of Drosophilaspecies was notpossible whendispersalbetweentemperatures threereplicatesofeach pair-wisecombinationofspecies wereestablishedin single,unconnectedcages at all fiveclinetemperatures (10, 15,20, 25 and 30?C). The cages were sampled weekly until one species was absent fromthreeconsecutivesamples. Second, the effectof interactionswas determinedin a system,

Table 2. Coefficients (and P values) forthe correlationsbetweenreal numbersof adult fliesand thosepredictedby sigmoid of theform) = k/(1+ e"-hN) relationships

D. mnelanogaster 1998 British Ecological Society Journal ofAniimcal Ecology,67, 600-612

D. sidnUians

g

D. slubobsciu7a

10?C

15?C

200C

25?C

300C

0.3244 (0.026) 0.4919 (0.001) 0.3044 (0.023)

0.1608 (0.297) 0.4422 (0.002) 0.4208 (0.003)

0.3042 (0.056) 0.6646 (0.001) 0.4809 (0.001)

0.5580 (0.001) 0.5386 (0.001) 0.3635 (0.021)

0.5211 (0.006) No fit No fit

604 Individualistic responsesto climate change

whichwas bothclinaland alloweddispersal,byassessing the distributionand abundance of Drosophila speciesin combination.Threetwo-speciesclineswere initiatedby adding 50 D. siniulans and 50 D. sUbobscura (equal sex ratiosin both species)to everycage in each cline.In addition,to ensurethattheoutcomes of interactionsin clineswere not greatlyinfluenced by the proportionof fliesused to initiatethem,six additionaltwo-speciesclineswereset up usingdifferentstarting proportionsofthetwospecies.Threewere startedby addingD. subobscurac to existingD. simiiulacnsclines at the rate of 10% D. subobscurato D. siniulans and threemore by adding D. siMnulains to existing D. stibobscur-a clines at the same rate. Although these clines started with differentproportionsof the two species therewas ultimatelyno difference betweenthem[repeated-measure significant ANOVA (RMA) F2,132 = 0 64, P = 0 528].The data from all three methods of establishmentwere therefore combinedin subsequentanalyses. Three-speciesclineswereinitiatedwithequal proportions of each species as different startingfrequencies did not influencethe eventualoutcome in two-speciesclines.The fourreplicatesweretherefore all startedby adding 50 fliesof all threespecies to everycage ina cline(400 fliesofeach speciespercline). wereexaminedusingthe Enemy-victim interactions Five four-speciesclineswere parasitoidL. boular-di. createdby addingadultwasps (sex ratio 1: 1) to cold three-speciesclines at the rate of 10% of the preadultsin each cage irrespective of existingDr-osophila species.The additionswere made 4 weeks afterthe three-speciesclines had been established.The data fromthese clines were also used to calculate temof thevictimassemblage, diversities perature-specific thethreeDrosophilac species. SIMULATED

GLOBAL

(J 1998British Ecological Society ofA7?nimal Journiial Ecology,67, 600-612

Results CAGE SERIES AND CLINES

In cage seriesD. melanogasterappeared most abundant,withD. subobscuraleastabundantand D. simulans in an intermediate position(Fig. 2). However,a two-wayANOVA revealed that there were no sig-

1500 (a) 1000

500

WARMING

The effectof global warmingon speciesassemblages in systemswithbothspeciesinteractions and dispersal betweentemperatures was examinedbyincreasingthe overall cline temperature.Hot clines were created fromthenineexistingtwo-speciesand thefivethreespeciescold clines(containingjust Drosophilaspecies withoutparasitoids),by settingtheincubatorsto the specifiedhot-clinetemperatures. DATA

wereinterpolatedusingthe mean of the two counts beforeand the two afterthe missingvalue (Norusis 1994). Parasitoidsmay affectthe diversityof the assemblages to whichtheirhostsbelong(LaSalle 1993),so the diversityof the Drosophilaassemblagewas calculatedforeach temperature fromthe weeklypopulationsof thethreeDrosophilaspeciesin four-species clines. Simpson'sindex was used as the measureof Drosophiladiversity in cold three-species clines(without parasitoids)and four-speciesclines (with parain abunsitoids)becauseitis sensitiveto thedifference dance betweenspeciesratherthan to the numberof species present(species richness)(Magurran 1988). to speciesrichnesswas notrequiredas richSensitivity in both ness was close to two at all temperatures cline types.Simpson'sindex also has low sensitivity theabilityto distinguish to samplesizewhileretaining in speciescomposition.Assemblagediverdifferences as themean diversity overweeks sitywas determined weremade at all 20-25 and separatedeterminations fourcold clinetemperatures.

ANALYSIS

Data fromcage series and clines were analysed as longitudinalstudies of individual types of experimental systemsusing a multivariateRMA design (Crowder& Hand 1990) withpolynomialcontrasts and unique sums of squares. The weeklypopulation wereused as within-subject factorsand sysestimzates temtype(e.g. seriesvs. dlinesor one-speciesdlinesvs. two- and three-species dlines)and temperatureas a between-subjectfactor. Occasional mzissingvalues

I

}1I

0

1500 (b)

500. (c) 1500

0

Il

12 13 Incubator

14

Fig. 2. Populations ofDrosoph1ila (a), D. siniumelanogaste7 (b) andD. subobscu7a (c)inone-species series(a) where fans5 all fourincubatorsare at 20?~Cand in one-speciesdlines(D)

wheretheincubators areat 10,15,20 and25?~C.Errorbars ? 1SE basedonmeansofall mneasures.

605 A.J.Davis et al.

nificantdifferencesbetween species (F 294= 2 77, P = 0 082) nor, emphatically,between incubators (F3,24= 0 182, P = 0 908). There was also no significantinteractionbetweenspecies and incubator (F6,24 = 0 301, P = 0 930). Thus, in series,none of the threespeciesshowedany trendacrossincubators. In contrast,in open one-speciesclines,wherethere was a temperature gradient,thepopulationsweresignificantly different fromthose in serieswherethere was no temperaturegradient between incubators (RMA F, 104 = 31 59, P < 0 001) afteraccountingfor the variancecaused by species and temperature.In addition,the populationsreachedby each speciesin each incubatoralong a serieswerenotcorrelatedwith the populationsin the same incubatorswhen these were part of a cline (D. iiielanogaster 1.12 = 0 37, P = 0 23; D. simulansS112= 0 18, P = 0 57; D. subobscurCa r12= 0 30, P = 0 46). Significantcorrelation wouldbe expectedifthetemperature gradientinclines had no effecton singlespecies populations.Within clines all threespecies are presentthroughoutthe available temperaturerange (Fig. 2). However, the at all speciespopulationsweresignificantly different temperaturesexcept 10?C (two-tailed t-test, all d > 1 96, P < 0 05) and each speciesreacheditshighest population densityat a differenttemperature (Fig. 2). The temperatureoptima (calculated as the mean of temperature weightedby thenumberof flies at each temperature) were 20-92?C (SD = 0 74) for D. nelanogasterand 19 0?C (0 39) and 16 2?C (0 65) forD. simulansand D. subobscura,respectively. DISPERSAL

(j 1998British Ecological Society ofAnim7al Jolurnal Ecology,67, 600-612

Dispersalin clines,whateveritsabsolutevalue,markedly alteredthe populationsachieved,as therewere differences betweenopen and blockedcold significant one-speciesclines independentof temperatureand and D. subobweeklyvariationforD. nielcanogaster scllr'a (RMA, respectively,F, 40 = 7 90, P = 0 008; F140 = 15-69,P < 0 001). The difference was not significantforD. sin1iulans(Fl 48 = 3 58, P = 0 064). In open clines,where dispersalwas possible, all three thecline, speciesoccurredand reproducedthroughout occupyingpartsof theclimaticrangewhere,without dispersal,theydied out or becameveryrare.Without becameveryrare,and ultidispersalD. melacnogaster Dromatelydied out at 10?C, as did D. simulains. a did notdie out at 10?C butbecame sop/lilasubobscur in closed extinctat 25?C, the highesttemperature, clines(Fig. 3). Preventingdispersal changed species abundance withinparticulartemperaturesas well as changing In closed clinesD. nielanogaster speciesdistributions. was moreabundantat 20 ?C and D. si-tiulacis at 25 ?C than theywerein open dlines(one-tailed(-test,correctedforunequalvariances,t= 4 22, d.f.= 178,P < 0 005; t = 3 67, d.f.= 178, P < 0 005, respectively). Theywerebothless abundant,however,at 15?C (t~=

1500 - (a) 1000

500

1500

.

(b)

e 500 - L

1500

(c)

1000.

-500.

10

15

20

25

Incubatortemperature

Fig.3. Comparison ofDrosophila (a),D. siniulmnelanogaster (c) populations in openonefan?s(b) andD. suwbobscuraz speciesclines(F) and closedone-species clines(D). The apparently negative population forD. suwbobscur1a at 25?C is causedbya minorbiasin thepopulation estimator at this temperature forverylowrealpopulationls Errorbars+ 1SE basedonmeansofallmleasures.

3 20, d.f.= 178, P < 0 005 and t = 4 31, d.f. = 178, P < 0 005) as was D. sini1ulanssat 100C (t = 4 56, d.f. = 178, P < 0 005). Inl contrast,D. subobscura populationsweresignificantly reducedat 25?0C(t = 401, d.f.=152, P 6 0, d.f. = 340, P < 0 005 for1525 ?C) (Fig. 4b). Interactionsalso reducedthe range of D. subobscurabecause it died out at 25 ?C in both two-and three-species clines(Fig. 4c). Its abundance in two-and three-species clineswas also significantly reducedcomparedto one-speciesclines throughout the remaining temperaturerange (all t > 3 00, d.f.= 214 or 376, P < 0.005) (Fig. 4c). The proportionalreductionwas greaterat 15?C thanat 10?C withtheresultthatthehighestpopulationsoccurred at 10?C insteadof at 15?C as was the case in onespecies clines. Species interactionsthus shiftedthe apparentoptimumof D. subobscuratowards10?C. The presenceof L. boulardisignificantly changed D. simulansand D. subobscurapopulations (RMA, respectively, F178= 18 02, P < 0 001; F180= 10 36, P = 0 002) butD. melanogaster was largelyunaffected (Fig. 5a). Drosophila siniulanspopulations,already clineswithoutparasitoids, verysmallin three-species were reducedeven further at 20?C (t = 515, d.f.= 460, P < 0 005) (Fig. 5b). In contrast,D. subobscura showed no significant change at 20 and 25?C but reachedverysignificantly higherpopulationsat 15'C (t = 5 12, d.f.= 460, P < 0 005) and at 10?C (t = 7 29, d.f.= 460, P < 0 005) whenwasps werepresent thanwhentheywerenot (Fig. 5c). These populations wereclose to thosereachedby D. subobscur-a whenit was on itsown (Fig. 2c). As a consequenceof thesechangesto Drosophila rangeand abundance,thepresenceof theparasitoid also increasedflydiversity because Simpson'sD was reducedat 10and 15'C (respectively, t= significantly 607, d.f. = 4, P < 005; t = 342, d.f. = 4, P < 005)

is also reflected in (Fig. 6). This increasein diversity the long-termcoexistenceof D. subobscuraand D. at 15?C in four-speciesclines, when nmelanogaster waspsarepresent,whereasD. subobscurais eventually excludedbyD. melanogaster at thesame temperature in three-species clines(Fig. 7).

607 A.J. Davis et al,

1500 1000

(a) -

500

InI""v

?'"1

(b)

1000 500

1000 (c) 0

10 -500

therealso was forD. simiiulanis (RMAF1,54 = 33 27, P < 0 001). However,simulatedwarmingof two-species clines,whereD. m1elanogasterwas absent,did not significantlyalter D. siniuitlans populations (RMA = = P 0 In 0 44, 513). two-species clinesD. sinrnuF1,30 lan1smaintaineda populationat 30?C but failedto do so in three-species clines. Drosophilasubobscuria populationswere significantly alteredin both twospeciesand three-species hot clinescomparedwithits populations in cold clines (RMA F,30 = 9 55, P = 0-004; F1,54 = 5 18, P = 0 027, respectively).In addition,as well as failingto establishpopulations at 30?C in eithertwo-or three-species hot clinesD. subobscurawas drivento negligiblelevels at 25 'C. Most markedly,however,D. subobscuraabundance was significantly higherat 15'C undersimulatedglobal warmingthanin cold clinesat thesame temperature,whetherin two-speciesor three-speciesclines

15

20

25

-

Incubatortemperature (t = 6.50, d.f. = 250, P < 0 005; t = 6 17, d.f. = 418, Fig. 5. ComparisonofDrosophilamiielaniogaster (a), D. sim11ulP < 0-005) (Fig. 8). Populations of D. sim11ulans in facns (b) and D. suibobscura (c) populationsin three-species in hot and cold at were two-species clines 15?C not, (U), andin without theparasitoid bolularcdi clines, Leptopilin1ac

four-species clines,withtheparasitoid (D). Theapparently forD. slubobscura at 25?C is due to a negative population minorbiasin thepopulation estimator at thistemperature forverylowrealpopulations. Errorbars + 1SE basedon meansofallmeasures.

1.5

however,significantly different (t = 0 68, d.f. = 250, P > 0 05) and D. nielanogaster and D. sinulains populations in hot three-species clines were significantly lower than those in similar cold clines (t = 2 86, d.f. = 418, P < 0 005; t = 3.08, d.f. = 418, P < 0 005;

Fig. 8). At 15'C, therefore, simulatedglobalwarming increasedD. subobscur)apopulationswhilstreducing those of its potentialcompetitors.This differential effectof global warminginvertedthe relativeabundance of the species at 15'C with D. subobscuca becomingdominantover allospecificsin hot clines, whereasin cold clinesit was least abundant at this temperature.

Discussion This study demonstratesthat dispersal and interactionsbetweenspeciescan modifyand disruptlinks betweentemperatureand species' local presenceor local abundance.These factorswill thusdistortpredictionsof rangeshiftsunderglobal warmingas it is 0 axiomaticin ecology that species existnot on their 10 15 20 25 own but in dynamicequilibriumwithothers,either Incubatortemperature horizontally(competition)or vertically(enemy-vicFig.6. Diversity (Simpson's D) oftheDrosophilca assemblage timinteractions) and thatlocal presenceand absence atallfourincubator inthree-species clines(l), temperatures are stronglyinfluencedby dispersalbetweenpoputheparasitoid without andinfour-specbouilard6ci, Leptopilin1ac lations(Hanski & Gilpin 1997). iesclines(D), withtheparasitoid. Errorbars+ 1SD. has long been Climate,particularlytemperature, acceptedas a dominantinfluenceon thedistribution and abundance of species (Messenger 1959; Coope 1977), an influenceexertedthrougheffectson fecSIMULATED GLOBAL WARMING undityand mortality.Temperatureis clearlyimporWhen global warmingwas simulatedby raisingthe tantin thisstudyas thereis no correlationbetween D)osophila speciesabundancein singlespeciesseries overallclinetemperatures by 5 ?C to createhotclines, and in dlinepopulations.Temperature distributionand abundance were both changed affects thethree (Fig. 8). There was a significant difference between species differently because theydifferin theirtempopulationsof D. melanogaster in hot and in cold peratureoptima. Drosoph1ila subobscura,whose recorded temperature three-species (RMA, F1,54 = P optima of 16 5 ?C (Moreteauetal. as dlines 9-79, =0-003) 0.5

?) 1998British Ecological Society ofAnimXcal Jour-nal Ecology,67, 600-612

608 Individualistic responsesto climatechaange

09

0.9 0.8

Y

oe

06

w

0.5",Z_ 0.7 0.6 0.4 0.3 0.2 0.1

0

0

5

10

20 30 15 25 Time sinceinitiation(weeks)

35

40

a (A) in threeFig.7. ComparisonofthesmoothedweeklyproportionsofadultDr-osophilaiiielaniogaster (0) and D. subobscur clines,withL. boulairdi parasitoids boulcorcdi parasitoids(closed symbols),and in four-species speciesclines,withoutLeptopilinia (open symbols).BothDrosophilaspeciespersistin thepresenceoftheparasitoidbut,in itsabsence,D. n7elanogasterultimately is omitted.) excludesD. subobsciura. (For clarity,data forthethirdmemberof theassemblage,D. simnudans,

1500 - (a) 1000 500

1500 - (b)

ilooo

:5 000-

Iooo

-

I

tL

(c)

500 : 0o

10

15

20

25

30

-500 Incubatortemperature (a), D. simu1zlFig.8. Comparisonof Drosophilaniielaniogalster laios(b) and D. subobscu7a(c) populationsin cold (V), and in hot (D) two-speciesclines;and in cold (E) andin hot(1) clines.The apparently negativepopulationforD. three-species subobscuraat 25?C is due to a minorbias in the population Error atthistemperature forverylowrealpopulations. estimator bars + 1SE basedon meansofall measures.

(9 1998 British Ecological Society JournalofAnim7al Ecology,67, 600-612

1997)or 19?C (Krimbas 1993)accordwiththesinglecoolspeciesclineoptimumofnear 15?C, is evidently adapted compared to D. melanogaster, temperate strainsof whichhave optimanear thatfoundin.the singlespeciesclinesin thisstudy(cf.David et al. 1983; Delpuech etal. 1995).As in thesinglespeciesclinesin thisstudy,D. simulans has an intermediate optimum (Cohet,Vouidibio& David 1980).Ifindividualphysiology was indeed the key determinantof species by eithercompetitionor dispersal, ranges,unaffected

it would be expectedthat thesethreespecieswould clinesin the orderthemselvesalong the temperature orderof theirtemperature optima,occurringat less abundancein neighbouring parts of the clines.This was not thecase in thepresentstudy. Dispersal substantiallymodifiesthe pictureand allows species, in the absence of interactionswith otherspecies,to occupylargerrangesthanphysiology or the rangesin blocked clines would suggest.The 'fundamental'niche (Hutchinson1958) in the single of each of the Drosophila dimensionof temperature species is thereforesmallerthan the realized niche whendispersalis possible;thatis,theDrosophilaform sinkpopulations(Pulliam 1988) in partsof thecline. An analogous situationoccurredforplantson sand dunes (Watkinson1985) and is a featureof sourcesink population dynamics. The dispersal rates betweentemperaturezones in clines in the present studyare higherthan theyare likelyto be formany organismsin nature,even though individualDrosophilacan disperseoverlargedistancesof 5-500km (Gressit& Nakata 1964;Jonesetal. 1981;Coyneetal. 1982). Nevertheless,this systemdemonstratesthe potential effectof long-distancedispersal such as occursin, forexample,the pest insectsNilaparvarta lugenisand Heliothisvirescens.Because dispersalcan evidentlybe veryimportant,should species' ranges shiftas a resultof climatechange,thuschangingthe spatialpatternof source-sink dynamics,speciesabundance and patternsof occurrencewill also change. Under thesecircumstances, 'null model', simplecliwill not accuratelypredictnew maticextrapolations, rangesand abundance. The degreeof inaccuracyis unknownbut would be greatestin specieswithsubstantialsinkpopulationsin partsof theircurrentdistribution. The resultsof single-cagepair-wiseinteractions indicatethattemperature influences theoutstrongly The data conformto thegeneral come ofcompetition. patternthat cool-adapted species are competitively

609

dominant at low temperaturesand warm-adapted species at hightemperatures, althoughtheydid not show thecommonlyobservedeliminationof D. mcelanogasterbyD. sim1tdlans at low temperatures (Moore 1952; Tantawy & Soliman 1967). On temperature clines, therefore,species with different temperature preferences shouldreplaceeach otherwithlittleoverlap. However, in two-speciesand in three-species clines this did not happen, because there is considerable species overlap maintainedby dispersal, whichopposescompetitive exclusion.Nevertheless, in theexperimental systemin thisstudytheintroduction ofcompetitors changesrangeand abundanceas competitorsare wellknownto do in nature(Connell 1961; Lawton & Hassell 1984).The effects differed between species,withD. stibobscwl)a abundancebeingreduced more than that of D. simiiilains in two-speciesclines, and D. simoulanssuffering morethaneitheroftheother speciesin three-species clines. Theory shows that synergismsbetweendispersal and interspecific interactions havethepowerto induce ininsectpopulations(Kareiva 1990; spatialpatterning Holmes etal. 1994) and thereis evidence that the spatial distributionof the westerntussock moth, by parasitism(Brodmann, Orgyiavetusa,is restricted Wilcox& Harrison1997).It is extremely unlikelythat all the speciesinvolvedin phenomenaof thesekinds will be affectedin the same way by climatechange, and existingspeciesassemblagesare thusbound to be disrupted.The degreeof disruptionwill depend on theintensity and thescale of theseeffects. Thus rangeand abundanceare strongly influenced both by the identityand the numberof competing species,and individualspeciescan be differently affecof othercompetitors in ted,dependingon theidentity the assemblage. Consequently, if each species respondsto climatechangein a different way,as the widespreadoccurrenceof non-analoguecommunities as suggestsis verylikelyfor organismsas different insects(Coope 1978, 1995), mammals (Graham & Lundelius1984;Grahametal. 1996)and plants(Davis 1986; Davis & Zabinski 1992), it must be expected that existingcombinationsof species will break up and new combinationswithdifferent relativeabundancewillformas speciesrespondand migratediffertheidentityand number entially.Inevitablytherefore of competitorsare likelyto change.The multiplenatureof rangeconstraintsis also evidentin data from a quite different source. Britishmacrolepidoptera occupy,on average,only0 37% of the geographical range of theirlarval food plants and in manycases muchlessthanthis(Quinn,Gaston & Roy 1997).The precisereasons forthis discrepancyare unclearbut mothsevidentlyrequiremore than just theirlarval food plant and have other,sometimessevere,constraints on their distribution.Under these Cirg 1998 British cumstances it is extremelyunlikelythat changes in Ecological Society nal ofAnim11al food plant and moth distributionswill be closely Jouri Ecology,67, 600-612 coupled. Moths cannlotOCCUrwithout their food

A.J. Davis et al.

plantsbutthistellsus nextto nothingabout a moth's actual distribution withintherangeof its food plant, eithernow or underglobal warming. Giventheimportanceofthesecompetitive phenomena, it is extremelyunlikelythat changes in range and abundancewillbe accuratelypredictedbyclimate mapping or single-speciesresponse norms alone. Again,themagnitudeof thediscrepancyis unknown; itwilldependon thenumberofcompetitors, theintensityof competitionand theextentto whichpotential showdifferential competitors rangechangesas climate changes(Graham 1992;Valentine& Jablonski1993; Coope 1995;Brubaker& McLachlan 1996). The introductionof a naturalenemyin the form of L. boulardialso has markedeffectson the three Drosophilac species.The increasein D. stibobscuraat 10?C is intriguing, itselfdoes not occur as L. boutlairdi at 10?C. The flymusttherefore be respondingto the parasitoid's suppressionof Drosophilaelsewherein theclineand thusa reductionin thenumbersofcompetitorsdispersing intothe10'C cages.The parasitoid thusfavoursD. subobsctira at lowertemperatures by fromtheother reducingthecompetition itexperiences D. mcelanogasteris two species.At hightemperatures favouredbecause, althoughthe parasitoidoviposits in D. nielanogaster, membersof theothertwo species thatit attacksare also killed.As a consequenceofthe on D. melanogaster and itsindirect wasp's local effects effectson D. subobscura,presenceof the wasp promotescoexistenceat 15'C. The presenceofD. subobsctiraat 15'C in a systemwhereit interactswithtwo otherspeciesis thusdetermined with byitsinteraction a natural enemyand not merelyby its individual indicatethatthewasp and physiology.These findings theDrosophilaspeciesareinvolvedincomplexindirect interactions (Holt & Lawton 1994). Similarcomplex interactionsare imputedfor two grapevine-feeding leafhoppersand a sharedparasitoid(Settle& Wilson betweentwo aphid spec1990) and are demonstrated ies and their natural enemies (Muller & Godfray 1997). Because all speciesareenmeshedin a webofnatural enemiesin theformof predators,parasitesor pathogens, such dynamicsare likelyto be frequent.It is veryunlikelythat all the interacting organismswill respondin preciselythe same way to climatechange and thus,in directconsequence,existingenemy-victiminteractions willbecomeuncoupledand newones established.This uncouplingor coupling of interactions by global climate change may have major impactson speciesrangesand abundance.It is once again evidentthat such impactswould override,or greatlymodify,changesarisingdirectlyfromspecies' individualresponsesto climatechange. This studyhas shownthatdispersal,horizontaland verticalinteractions have major effects on theranges and abundance of the speciesmakingup an assemblage. They modifyor disruptthe linksbetweenclimate and species' local presenceor local abundance

610 Individualistic responses to climate change

References

Environment (eds R. experiment.Insects in a Chlanginig Harrington& N. E. Stork),pp. 50-59. Academic Press, London. Boorman,S.A. & Levitt,P.R. (1973) Group selectionon the boundaryof a stable population. TheoreticalPopulation Biology,4, 85-128. Brodmann,P.A., Wilcox,C.V. & Harrison,S. (1997) Mobile parasitoidsmay restrictthe spatial spread of an insect nal ofAniimal outbreak.Jouri Ecology,66, 65-72. Brubaker,L.B. & McLachlan, J.S. (1996) Landscape diverclimatechange sityand vegetationresponseto long-term USA. in theeasternOlympicPeninsula,PacificNorthwest, anidTe rrestr ial Ecosystenms (eds B. Walker Global Chcanlge Press, pp. 184-203.CambridgeUniversity & W. Steffan), Cambridge. Brussard,P. (1984) Geographicpatternsand environmental gradients: the central-marginalmodel in Drosophila Reviewof Ecologyand Systeniatics, revisited.Annuitial 15, 25-64. Cammell, M.E. & Knight,J.D. (1992) Effectsof climate change on the population dynamics of crop pests. inEcologicalResearch,22, 117-12. Advanices Carter,T.R., Parry,M.L. & Porter,J.H. (1991) Climate changeand futureagroclimaticpotentialin Europe.Interniational Journ1al of Climiatology, 11, 251-269. Carton,Y., Bouletreau,M., van Alphen,J.J.M.& van Lenteren,J.C.(1986) The Drosophilaparasiticwasps. TheGe;ieticsan1dBiologyof Drosophila,Vol. 3e (eds M. Ashburner, H. Carson & J. N. Thompson), pp. 3471394. AcademicPress,London. Carton, Y., Haouas, S., Marrakchi,M. & Hochberg,M. (1991) Two competingparasitoidspeciescoexistin sympatry.Oikos,60, 222-230. Caughley,G., Short,J.,Grigg,G.C. & Nix,H. (1987) KangaJournalof roos and climate:an analysisof distribution. AniinalEcology,56, 751-761. Cohet,Y., Vouidibio,J. & David, J.R. (1980) Thermaltolerance and geographic distribution:a comparison of cosmopolitanand tropicalendemicDrosophila species. Journ'Mal of therm11al Biology,5, 69-74. Connell,J.H. (1961) The influenceofcompetitionand other factorson the distributionof the barnacle Chthanmalus stellatucs. Ecology,40, 49-78. Coope, G.R. (1977) Fossil coleopteranassemblagesas sensitiveindicatorsof climaticchangeduringthe Devensian (last) cold stage. PhilosophicalTransactionsof theRoyal SocietyofLondon,SeriesB., 280, 313-340. Coope, G.R. (1978) ConstancyofinsectspeciesversusinconDiver-sity of Insect stancyof Quaternaryenvironments. Faulnas (eds L. A. Mound & N. Waloff),pp. 176-187. Symposiaof theRoyal EntomologicalSocietyof London PublicationsLtd, Oxford. No. 9. BlackwellScientific of quaternaryclimaticchanCoope, G.R. (1995) The effects ges on insectpopulations:lessonsfromthepast. Insectsin a Chaniginig Enivironmenit (eds R. Harrington& N. E. Stork),pp. 30-48. AcademicPress,London. Coyne, J.A., Boussy, I.A., Prout,T., Bryant,S.H., Jones, J.S. & Moore, J.A. (1982) Long-distancemigrationof Drosophila.Amiierican Naturalist,119, 589-595. Crowder,M.J. & Hand, D.J. (1990) Analysisof Repeated Measur-es.Chapman & Hall, London. David, J.R.,Allemand,R., van Herrewege,J. & Cohet,Y. (1983) Ecophysiology: abiotic factors. Genetics and Biologyof Drosophila, Vol. 3a (eds M. Ashburner,H. Carson & J.N. Thompson),pp. 106-170.AcademicPress,

Beerling,D.J. (1993) The impact of temperatureon the northern distribution limitsof theintroducedspeciesFallopia japonica and Imspatienis glanduliferain north-west Europe. Jour-nal ofBiogeogr aphy,20, 45-53. Bennetts,D.A. (1995) The Hadley Centretransientclimate

Davis, A.J.,Varley,M.E., Baker,R.H.A. & Hardy,I.C.W. (1996) ParasitoidsofDrxosophilain theBritishIsles. En1to115,1-13. niolo0gist, Davis, M.B. (1981) Quaternaryhistoryand the stabilityof ForestSucecession, ConceptsandAppliforest communities.

and willthusdistortpredictionsof rangeshiftsunder global warming(Pacala & Hurtt 1993). Dispersal allows species to maintainpopulations away from theirphysiologicaloptimaand to interactwithspecies comInterspecific climaticpreferences. withdifferent petitiontends to reduce species' abundance but its so thata and species-specific effects are temperaturespeciesmaybe reducedin partof its rangebut not in others.Interspecific competitionmay also shiftspecies' optimaand reducespeciesranges.Naturalenembecausetheycan ies producemorecomplicatedeffects reduceabundance in some parts of a species' range and yet increaseit in othersthroughindirectinteractions withcompetitors.In consequence,a system includingspeciesinteractionsand dispersalcan produce substantialunexpectedand counter-intuitive effectseven in a simplelaboratorysysteminvolving no more than fourinteractingspecies. There is no reason to believe that similar'unexpected'changes in range and abundance will not occur in natural populations,providingonlythatcurrentdistributions are maintainedand modifiedby a combinationof climate,dispersal(source-sink)dynamicsand interactionswithotherspecies,and that speciesrespond to climatechange. idiosyncratically to determine ofchanIn attempting thelikelyeffects ges in climate,or indeedof clinallyvaryingfactorsin we must general,on speciesrangesand distributions, startaskinghowimportant dispersaland speciesinteractions are in natural and agriculturalecosystems, how intensetheyare and on whatscalestheyoperate. Althoughit is likelythatthepole-wardmovementof positedas thekeyprediction speciesranges,frequently of global warming(Peters& Darling 1985; Davis & Zabinski1992;Woodward 1992;Parmesan1996),will on species, remainbroadlycorrect,thedetailedeffects includingpests,disease vectorsand species of conservationconcern,will be problematicand some of these problems may be expensive. Models incorporatingdispersal and species interactionswill be requiredfor adequate predictionsof the potentially seriousappliedconsequencesof global warming.

Acknowledgements We thankP. Nicholson forstatisticaladvice and M. Wilbrahamforeditorialassistance.The researchwas supportedby BBSRC grant24/GER00620 to ProfessorsJ.H. Lawton& B. Shorrocksand is partof the BBSRC Biological Adaptationsto Global EnvironmentalChangeprogramme.

(9 1998 British Ecological Society Joutr-nal ofAnimal Ecology,67, 600-612

London.

611 A.J. Davis et al.

(9 1998 British Ecological Society JournclofAnimal Ecology,67, 600-612

ccation1 (eds D. C. West,H. H. Shugart& D. B. Botkin), Kohlmann,B., Nix, H. & Shaw,D.D. (1988) Environmental pp. 132-153.SpringerVerlag,New York. limitsofchromosomaltaxa predictionsand distributional Davis, M.B. (1986) Climateinstability, timelags, and comin theAustraliangrasshopperCaclediac (F.). Oecocacptival munitydisequilibria.Commiunity Ecology(eds J.Diamond logia,75, 483-493. & T. J.Case), pp. 269-284. Harper& Row, New York. Kraaijeveld,A.R. & vanAlphen,J.J.M.(1995) Geographical Davis, M.B. & Zabinski,C. (1992) Changes in geographic variation in encapsulation ability of Drosophilacimielrangeresulting fromgreenhousewarming:effects on biodilarvaeand evidenceforparasitoid-specific an71ogaster comversityin forests.Global l47arming canld BiologicalDilversity ponents.EvoluttionarY Ecology,9, 10-17. (eds R. L. Peters& T. E. Lovejoy), pp. 297-308. Yale Krimbas,C.B. (1993) Drosophila subobscuraBiology,Ge;iUniversity Press,New Haven. eticsan71d InversioniPolymorphisnm. VerlagDr. Kovac, HamDelpuech,J.M.,Moreteau,B., Chiche,J.,Pla, E., Vouidibio, burg. J.& David, J.R. (1995) Phenotypicplasticityand reaction LaSalle, J. (1993) Parasitichymenoptera: biologicalcontrol normsintemperate and tropicalpopulationsofDrosophila and biodiversity.Hy)mnenoptera ans1dBiodiLersitV (eds J. mnelaniogaster--ovariansize and developmentaltemperaLaSalle & I. D. Gauld), pp. 197-215.CAB International, ture.Evolution, 49, 670-675. Wallingford. Gilpin,M. & Hanski, I. (1991) Metalpoplulactioni Dynwiaimfics.Latreille,P.A. (1819) MAmoires surSdiverssuijetsle i'histoire EmpiricalaniidThieor eticacl AcademicPress, Inivestigactionis. clesinisectes, n1alturselle dlegeographieancienniie et dlec/lironLondon. ologie.A Berlin,Paris. Graham,R.W. (1992) Late Pleistocenefaunalchangesas a Lawton, J.H. (1995) The responsesof insectsto climatic of greenhousewarmingon guideto understanding effects change. Inisectsin acChacnginig Envi0ronment (eds R. HarthemammalianfaunaofNorthAmerica.Global PWarming rington& N. E. Stork),pp. 3-26. AcademicPress,London. aindBiologicalDiversity) (eds R. L. Peters& T. E. Lovejoy), Lawton, J.H. & Hassell, M.P. (1984) Interspecific compp. 76-87. Yale University Press,New Haven. petition in insects. Ecological Entoinologvn (eds C. B. Huffaker & R. L. Rabb), pp. 451-420.JohnWiley& Sons, Graham, R.W. & Lundelius,E.L. (1984) Co-evolutionary Chichester. disequilibria and Pleistocene extinctions. Quiaterniary Extinctions.al Prehistoric Rev7oluttion (eds P. S. Martin& aniditsMeasurveMagurran,A.E. (1988) EcologicalDiversity R. G. Klein), pp. 223-249. Universityof Arizona Press, inienit. Croom Helm Ltd, London. Tucson. Messenger,P.S. (1959) Bioclimaticstudies with insects. Reviewi Graham,R.W., Lundelius,E.L., Graham,M.A., Schroeder, Annui1lal ofEnitom0ology, 4, 183-206. nielE.K., Toomey, E.S., Anderson, E., Barnosky,A.D., Moore, J.A. (1952) CompetitionbetweenDrosophlilal and D. simulllanis. Burns, J.A., Churcher,C.S., Grayson, D.K., Guthrie, I. Populationcage experiments. an0ogaster R.D., Harington,C.R., Jefferson, G.T., Martin, L.D., Evoluttion, 6, 407-420. Moreteau,B., Morin,J.-P.,Gibert,P., Petavy,G., Pla, E. McDonald, H.G., Morlan, R.E., Semken,H.A., Webb, & David, J.R. (1997) Evolutionaryshiftof continuous S.D., Werdelin, L. & Wilson, M.C. (1996) Spatial responsesof mammalsto late quaternaryenvironmental nonlinearreactionnormsin a cold-adaptedDrosophila fluctuations. Rendusde l'Acadenfie des Scienices, Serie Scienice,272, 1601-1606. species.Comiptes Gressit,J.L.& Nakata, S. (1964)Trappingofairborneinsects III, Scienicesdlela Vie-LifeSciences,320, 833-841. on shipsin thePacific.Part 5. Paccific Mfiller,C.B. & Godfray,H.C.J. (1997) Apparent comIiisects,6, 274-282. Hanski,I.A. & Gilpin,M.E. (1997) Metalpoplulationi petitionbetweentwo aphid species. Journlalof Animial Biolog). aniidEtolutioni. Academic Press, LonEcology,Genietics, Ecology,66, 57-64. don. Nix,H. (1986) A biogeographicanalysisofAustralianelapid A.A. & Blows, M.W. (1994) Species borders: snakes. Atlalsof'Aulstraliani Hoffim1ann, ElacpidSniakes(ed. R. Longin Ecolecologicaland evolutionaryperspectives.Trendcls more),pp. 4-15. Bureau of Flora and Fauna, Canberra. ogyans1d Nordlander,G. (1980) Revision of the genus Leptopiliia Evoluttion, 9, 223-227. Forster, 1869, with notes on the status of some other Holmes,E.E., Lewis,M.A., Banks,J.E. & Veit,R.R. (1994) Partial differential equations in ecology: spatial intergenera (Hymenoptera,Cynipoidea: Eucoilidae). Enitoactionsand populationdynamics.Ecologjy,75, 17-29. n0iologica Scandinav7ica, 11, 428-453. 6.1. Chicago, Norusis,M.J.(1994) SPSS Ad,ancedStaltistics Holt, R.D. & Lawton, J.H. (1994) The ecological conseofEcolSPSS Inc. Rev7ieiw quencesofsharednaturalenemies.Annuliacl Overpeck,J.T., Webb, R.S. & Webb, T. (1992) Mapping ogy7an1dS)steinatics,25, 495-520. easternNorth Americanvegetationchange of the past Hutchinson,G.E. (1958) Concludingremarks.Cold Sprin1g oniQluan71titaltive 18ka: Non-analoguesand the future.Geology,20, 1071HarsbolurSyvinposilumn Biology,22, 415-427. E.P. & Jeffree, C.E. (1994) Temperatureand the 1074. Jeffree, of species.Funs1ctionlacl EcolG.C. (1993) Terrestrial Pacala, S.W. & Hturtt, vegetationand biogeographicaldistributions climatechange:integrating modelsand experiments. Biotic ogy,8, 640-650. E.P. & Jeffree, acctionis GlobaclChans1ge C.E. (1996) Redistributionof the Iiiter Jeffree, ans1d (eds P. M. Kareiva, J. G. Kingsolver& R. B. Huey), pp. 57-74. Sinauer,Sunderpotentialgeographicranges of Mistletoeand Colorado Beetle in Europe in response to the temperaturecoi11land,MA. Paine, R.T. (1966) Food web complexityand speciesdiverponent of climatechange. Fuln1ctionlacl Ecology,10, 562577. sity.AinericanNactursalist, 100,65-75. Parmesan,C. (1996) Climateand species'range.Natture, 382, Jones,J.S., Bryant,S.H., Lewontin,R.C., Moore, J.A. & 765-766. Prout,T. (1981) Gene flowand thegeographicdistribution of a molecular polymorphismin Drosophilalpseiudlo- Parry,M.L. (1990) Clim?1ate changean1d WVorld Agricudlture. obscuri ac.Genietics, Earthscan,London. 98, 157-178. Patterson,J.T. (1943) Stuldiesin thegenietics of Drosophila. Kareiva,P.M. (1990) Populationdynamicsin spatiallycomTra,isIII. The Drosoplhilidace plex environments: theoryand data. Philosophicacl of the SouthwiYest. Universityof Texas Publications,No. 4313. oftheRoyallSocietyofLon1don, Series B, 330,175acetions1 Peters,R.L. (1992) Conservationof biologicaldiversityin 190. thefaceofclimaticchange.GloballWXarmini7g Kingsolver,J.G. (1989) Weatherand thepopulationdynamalndBiologicall (eds R. L. Peters& T. E. Lovejoy), pp. 16-30. ics of insects:integratingphysiologicaland population Diversit.y ecology.Physiologicall Zoology,62,314-334. Yale University Press,New Haven.

612 Individualistic responsesto climatechange

? 1998British Ecological Society JournalofAnimal Ecology,67, 600-612

Peters,R.L. & Darling,J.D.S. (1985) The greenhouseeffect and naturereserves.Bioscience,35, 707-717. Porter,J. (1995) The effectsof climatechange on the agriculturalenvironment forcrop insectpestswithparticular referenceto the European corn borer and grain maize. Insectsin a ChangingEnvironm?1enit (eds R. Harrington& N. E. Stork),pp. 93-123. AcademicPress,London. Porter,J.H.,Parry,M.L. & Carter,T.R. (1991) The potential effects ofclimatechangeon agricultural insectpests.Agriculturhal and For-est Meteor-ology, 57, 221-240. Price,P.W. (1992) Plant resourcesas the mechanisticbasis for insect herbivorepopulation dynamics. Effectsof Resour7 ce Distribuitioni on Animnal-Plant Initer actionis(eds M. D. Hunter,T. Ohgushi& P. W. Price),pp. 139-173. AcademicPress,London. Pulliam,H.R. (1988) Sources, sinks and populationregulation.Amnerican Naturalist,132,652-661. Quinn,R.M., Gaston,K.J. & Roy, D.B. (1997) Coincidence betweenconsumerand hostoccurrence:macrolepidoptera in Britain.EcologicalEntomology, 22, 197-208. Rodriguez,J.,Jordano,D. & FernandezHaegar, J. (1994) in a butterfly-host Spatial heterogeneity plantinteraction. JournalofAnimiial Ecology,63, 31-38. Rogers, D.J. (1979) Tsetse population dynamicsand distribution:a new analyticalapproach. Jorn-nal of Aniinal Ecology,48, 825-849. Rogers,D.J. & Randolph,S.E. (1993) Distributionof tsetse and ticksin Africa:past,presentand future.Parasitologv Today,9, 266-271. Roy, K., Valentine,J.W.,Jablonski,D. & Kidwell, S.M. and timeaveragingin (1996) Scales of climaticvariability Pleistocenebiotas:imliplications forecologyand evolution. Treinds in Ecologyand Evolution,11, 458-463. limitsfor Scott,D. & Poynter,M. (1991) Uppertemperature troutin New Zealand and climatechange.Hydrobiologia, 222, 147-151. Settle,W.H. & Wilson, L.T. (1990) Invasion by the varicomnegatedleafhopperand bioticinteractions: parasitism,

petitionand apparent competition.Ecology, 71, 14611470. Shorrocks,B. (1972) Drosophila.Ginn & Co, London. Sutherst, R.W. (1990) Impactofclimatechangeon pestsand diseasesin Australia.Sear-ch, 21, 230-232. Sutherst,R.W. & Maywald, G.F. (1985) A computerised systemformatchingclimatesin ecology.Agriculture, Ecoa1d Environnment, 13, 281-299. systenms Sutherst,R.W., Maywald,G.F. & Skarrat,D.B. (1995) Predictinginsectdistributions in a changedclimate.Insectsin a Chanigiing Enivir0onment (eds R. Harrington& N. E. Stork),pp. 60-93. AcademicPress,London. Tantawy,A.0. & Soliman,M.H. (1967) Studieson natural populationsofDrosophila.VI. CompetitionbetweenDroand D. sinmtlains. sophilamiielanogaster Evolutioni, 21, 3440. Valentine,J.W.& Jablonski,D. (1993) Fossil communities: compositionalvariation at many time scales. Species in EcologicalComnm1unities Diver-sit:y (eds R. E. Ricklefs& D. Schluter),pp. 341-349. Universityof Chicago Press, Chicago. Walter,G.H. & Patterson,H.E.H. (1994) The implications of palaeontologicalevidence for theoriesof ecological Journalof communitiesand species richness.Austr-alian Ecology,19, 241-250. Watkinson,A.R. (1985) On the abundanceof plantsalong an environmental gradient.Journalof Ecology,73, 569578. Wheeler,M.R. (1981) The Drosophilidae,a taxonomicoverview.Th1eGeneticsand BiologyofDrosophila,Vol. 3a (eds M. Ashburner,H. Carson & J.N. Thompson),pp. 1-98. AcademicPress,London. of climateon Woodward,F.I. (1992) A reviewof theeffects vegetation:ranges,competitionand composition.Global Wariningand BiologicalDiversity(eds R. L. Peters& T. E. Lovejoy), pp. 105-123. Yale UniversityPress, New Haven. Received11 Jul:y received24 October1997 1997; r-evision