increased catecholamine metabolism) inthe central nervous system (10), thehippocampus could be one of the sitesof this restoration process.The electrical activity of the hip-

pocampus is unusual. While the rest ofthe brain shows a desynchronized EEGpattern during attention, learning, andparadoxical sleep, the hippocampus dis-plays a synchronized electrical activityof four to seven cycles per second (thetarhythm) (11). It is suggested that thehippocampus, through synchronizing in-fluences on the neocortical and subcorti-cal structures, counteracts the effects ex-erted by the ascending reticular-acti-vating system, and that the functional in-terplay of the hippocampus and reticularsystem seems to be of importance withregard to the rhythm of sleep and wake-fulness (12). Spontaneous activity of thehippocampal pyramidal cells in cats withpermanently implanted electrodes sharp-ly decreased in SWS (13), a phenomenonobserved in other brain areas (14). It maybe that the electrical activity of specificbrain structures during SWS is keyed tolocal chemical processes. Our data re-garding the chemical changes in the hip-pocampus during SWS indicate that the5-HT and DA systems become activatedat this time. This is of interest becausecurrent hypotheses propose an increasedactivity in the 5-HT system of the raphenuclei during SWS and an increased ac-tivity in the catecholamine system of thenucleus locus coeruleus during para-doxical sleep (3). Because of the recipro-cal connection between limbic-forebrainstructures (including the hippocampus)and the raphe nuclei, the latter are poten-tially modulated by limbic-forebrainmechanisms (15). When 80 to 90 percentof the raphe system is destroyed, ani-mals enter a state of permanent arousalthat lasts 3 to 4 days; SWS returns par-tially within a 3-week period (2), and anear normal sleep profile (16) appears byday 30. This indicates that other brainsites functionally compensate for theloss of 5-HT neurons in the raphe. Fur-thermore, since hippocampectomy wasfound to reduce significantly both SWSand paradoxical sleep, the hippocampushas been implicated in the facilitation ofboth SWS and paradoxical sleep (17). Al-though hibernation differs from normalsleep in many respects (18), it has beenreported that when animals are enteringhibemation, the concentration of 5-HTincreases in the hippocampus severaltimes more than it does in other brainareas (19). Also, when 5-hydroxytrypto-phan, an immediate precursor of 5-HT,is administered to rats, the highest con-centration of 5-HT is found in the hip-10 SEPTEMBER 1976

pocampus (20); when administered torabbits, 5-hydroxytryptophan results inthe most marked rate of increase in 5-HTbeing found in the hippocampus (21). Allthese findings point to the importance ofa serotonergic mechanism in the hip-pocampus, and a possible role of thisarea in SWS. The specific increases inthe metabolism of 5-HT and the concen-tration of DA in the hippocampus duringSWS indicate that the hippocampus func-tions as a subsidiary sleep structure tothe raphe system and the nucleus locuscoeruleus in the brainstem. We also sug-gest that the obtained decrease in DA me-tabolism in the striatum and thalamusduring SWS may be related to the sleep-generating mechanisms.

RUkICA KOVAtEVI1*MIODRAG RADULOVAtKIt

Department ofPharmacology,College of Medicine,University of Illinois at the MedicalCenter, Chicago 60612

References and Notes

1. A. Dahlstrom and K. Fuxe Acta Physiol.Scand. 62, Suppl. 232 (1964).

2. M. Jouvet, P. Bobillier, J. F. Pujol, J. Renault,C. R. Seances Soc. Biol. Pars 160, 2343 (1966).

3. M. Jouvet, Science 163, 32 (1969).4. Since dura mater contains pain receptors, re-

moval of the dura prior to the expenmentalprocedure was an important part of the tech-nique. It enabled painless cutting of the braintissue and ablation of the left hemisphere withthe least disturbance to the animal.

5. The striatum-thalamus included the putamen,globus pailidus, tail of caudate nucleus, claus-trum, and thalamus.

6. C. C. Chang, Int. J. Neuropharmacol. 3, 643(1964); F. A. Gerbode and M. B. Bowers, J.Neurochem. 15, 1053 (1968); J. Korf and T.Valkenburgh-Sikkema, Clin. Chim. Acta 26,301 (1969); A. S. Welch and B. L. Welch, Anal.Biochem. 30, 161 (1969); H. S. Alpers and H. E.

As human destruction of remainingnatural habitats accelerates, biologistshave felt intuitively that most existingwildlife refuges are too small to avertextinctions of numerous species. How-ever, because there has been no firmbasis for even approximately predictingextinctions in refuges, biologists havehad difficulty convincing governmentplanners faced with conflicting land-usepressures of the need for large refuges.Recently several workers have recog-nized that a predictive understanding ofextinction might be obtained from islandbiogeography, since refuges of naturalhabitat in a sea of human-altered environ-ment behave as islands for species depen-dent on natural habitat (1-6). All theseinvestigators attempting to understandimplications of the "island dilemma" for

Himwich, J. Pharmacol. Exp. Ther. 130, 531(1972).

7. R. L. Buckingham and M. Radulovadki, BrainRes. 99, 440 (1975).

8. A. K. Sinha, S. Henriksen, W. C. Dement, J. D.Barchas, Am. J. Physiol. 224, 381 (1973).

9. B. E. Jones, P. Bobillier, C. Pin, M. Jouvet,Brain Res. 58, 157 (1973). i

10. E. L. Hartmann, Int. Psychiat. din. 7, 38(1970); W. C. Stem and E. L. Hartmann, Proc.Am. Psychol. Assoc. 7, 308 (1971).

11. E. Grastyan, K. Lissak, I. Madarasz, H. Don-hoffer, Electroencephalogr. Clin. Neurophysiol.11, 409 (1959); W. R. Adey, D. 0. Walter, C. E.Hendrix, Exp. Neurol. 7, 282 (1961); E. N.Sokolov, Annu. Rev. Physiol. 25, 545 (1963); M.Radulovacki and W. R. Adey, Exp. Neurol. 12,68 (1965); M. Jouvet, Prog. Brain Res. 18, 20(1965); A. Routtenberg, Psychol. Rev. 73, 481(1966); W. R. Adey, Prog. Brain Res. 27, 228(1967).

12. P. L. Parmeggiani, Prog. Brain Res. 27, 413(1967).

13. H. Noda, S. Mahonar, W. R. Adey, Exp. Neu-rol. 24, 217 (1969).

14. E. V. Evarts, Fed. Proc. Fed. Am. Soc. Exp.Biol. 19, 828 (1960); P. R. Huttenchlocher, J.Neurophysiol. 24, 451 (1961); E. V. Evarts, ibid.27, 152 (1964); E. Bizzi, 0. Pompeiano, I. So-mogyi, Arch. Ital. Biol. 102, 308 (1964); H.Sakamura, Jpn. J. Physiol. 18, 23 (1968).

15. P. J. Morgane and W. C. Stem, in Serotonin andBehavior, J. Barchas and E. Usdin, Eds. (Aca-demic Press, New York, 1973), pp. 427-442.

16. P. J. Morgane, in Sleep and the Maturing Ner-vous System, C. D. Clemente, D. P. Purpura, F.E. Mayer, Eds. (Academic Press, New York,1972), pp. 141-162.

17. C. Kim, H. Choi, C. C. Kim, J. K. Kim, M. S.Kim, H. J. Park, B. T. Ahn, Electroencepha-logr. Clin. Neurophysiol. 38, 235 (1975).

18. L. M. N. Bach, in Hibernation and Hypother-mia, Perspectives and Challenges, F. E. South,J. P. Hannon, J. R. Wiflis, E. T. Pengeiley, N.R. Alpert, Eds. (Elsevier, Amsterdam, 1972),pp. 535-550.

19. N. K. Popova and N. N. Voitenko, Dokl. Akad.Nauk SSSR 218, 1488 (1974).

20. F. Okada, Y. Saito, T. Fujeida, I. Yamashita,Nature (London) 238, 355 (1972).

21. E. Costa and F. Rinaldi, Am. J. Physiol. 194,214 (1958).

22. W. J. Dixon and F. J. Massey, Jr., Introductionto StatisticalAnalysis (McGraw-Hill, New York,1969), p. 119.

23. This work was supported by PHS NS 10921.* Present address: Institute of Physiology, Facul-

ty of Medicine, Belgrade, Yugoslavia.t Reprint requests should be sent to M.R.4 March 1976; revised 4 June 1976

conservation strategy have concludedthat some large refuges are essential tominimize extinction rates and to ensurecertain species any chance of survival atall. These conclusions are based not onlypn studies of oceanic islands but also ofhabitat "islands" on mainlands, as wellas of refuges themselves.

Simberloff and Abele (7) argue thatthese applications of biogeographic theo-ry to conservation practice are pre-mature and are based on insufficientlyvalidated theory and possibly also onidiosyncratic results. These authorsshow that, given certain assumptions,several small refuges may contain morespecies than a single large refuge ofequivalent area. Their reasoning fromtheir assumptions is correct but mini-mizes or ignores much more important

1027

Island Biogeography and Conservation:Strategy and Limitations

on January 24, 2020

http://science.sciencemag.org/

Dow

nloaded from

conservation problems. Because thoseindifferent to biological conservationmay seize on Simberloff and Abele'sreport as scientific evidence that largerefuges are not needed, it is important tounderstand the flaws in their reasoning.Human activities threaten some spe-

cies and habitats more than others. Hu-mans are preferentially destroying somehabitats (for example, primary tropicalrain forest) and creating others (for ex-ample, roadsides and pastures). In addi-tion, species of high dispersal ability andhigh reproductive potential, living in suc-cessional habitats, can survive human-related environmental changes much bet-ter than can sedentary species of lowreproductive potential that are confinedto more mature habitats. Thus, con-servation strategy should not treat allspecies as equal but must focus on spe-cies and habitats threatened by humanactivities. What are the area require-ments and dispersal abilities of theseextinction-prone species?

First, consider area requirements ofspecies that can disperse among islandsor habitat patches. Despite their abilityto disperse, such species are often foundto be confined to islands or patches muchlarger than the territory size of a singlepair [see (5) for summary]. For example,minimum area requirements of popu-lations of those southwest Pacific landbird species that can colonize islandsdy4rwater range up to thousands ofsctdare kilometers for species whose ter-ritories are measured in hectares (3).Iguanid lizard populations of the Baha-mas are confined to islands large enoughto support about 100 lizards (8). Similarexamples of minimum area requirementshave been reported for North Americanants, North American birds, and Britishbirds (9). These requirements result fromseveral factors (3, 5). (i) Some habitatsexist only on larger islands or patches;(ii) species with seasonally or spatiallypatchy food supplies must integrate re-sources over large areas; (iii) species thatlive at low densities, and hence oftenbecome extinct on small islands but rare-ly recolonize, have low probability ofoccurrence at equilibrium except onlarge islands; and (iv) "hot spots" oflocally high resource production may beimportant hedges against extinction butmay constitute only a small fraction ofbreeding territories.

For species capable of dispersal be-tween "islands," extinction of a popu-lation in one refuge may possibly bereversed by colonization from anotherrefuge. The island dilemma is posed inmore acute form by species that are un-able or unwilling to disperse across wa-

1(28

ter or alien habitat. Such species includenot only flightless organisms but alsobirds that are strong fliers. For example,many Californian bird species have nev-er been observed on islands or desertoases 20 km from their breeding sites(10); 302 of the 513 breeding land birdspecies of New Guinea have never beenobserved on a single oceanic island, noteven large islands 8 km from NewGuinea (4); and 55 of the 127 breedingland bird species of New Britain neverappeared, even as vagrants, on an 8-km2island 6 km offshore during several dec-ades' residence by Meyer, a keen observ-er (3). Since there are no immigrations toreverse extinction on an island or habitatpatch for such species, minimum arearequriements are considerably largerthan for species capable of dispersal.Thus, New Guinea is surrounded bymany "land-bridge islands" that formedpart of New Guinea during low sea levelof Pleistocene times up to 10,000 yearsago. At the present time, 32 of the 134New Guinea lowland bird species that donot colonize overwater have disappearedfrom all land-bridge islands, even ones aslarge as 8000 km2. These extinction-prone species with large area require-ments include some of the most dis-tinctive New Guinea bird species, suchas the vulturine parrot Psittrichas ful-gidus, harpy eagle Harpyopsis novae-guineae, and shovel-billed kingfisher Cly-toceyx rex (4). Yet few proposed refugesexceed 8000 km2 in area.

Similar patterns of differential post-Pleistocene extinction on real land-bridge islands in the ocean, or on virtualones in seas of alien habitat, have beendescribed for west Australian macropodmarsupials, southeast Australian marsu-pials and rodents, North American mon-tane mammals, neotropical birds, Bis-marck Archipelago birds, Solomon Is-land birds, and Australian lizards (2-5,11). In all of these studies, most bird andmammal species incapable of interislanddispersal were found to disappear fromall islands smaller than a few hundredsquare kilometers, and some species dis-appeared even from all islands of manythousand square kilometers. While thesepatterns are the product of populationfluctuations for about 10,000 years, stud-ies in this century on many New Zealandforest reserves (6) and on Panama's Bar-ro Colorado reserve (2) show that manyextinctions occur within a few decades,especially in smaller refuges.As a result of this differential suscepti-

bility of species to extinction in isolatedpopulations, small refuges or islandsmainly lose the sedentary species of ma-ture habitats that are most threatened by

human activities, and retain the rapidlydispersing successional and edge speciesthat need no protection. For instance,small forest reserves in New Zealandgradually lose all bird populations be-longing to old endemic families and re-tain a standard quota of birds that arealso widespread in suburban gardens,mostly species that recently immigratedor were introduced by Europeans toNew Zealand (6).

Simberloff and Abele (7) suggest sev-eral reasons that they believe argue forsmall refuges under some circumstances.

1) Their main argument is that, de-pending on species pool size and relativeareas of refuges, several small refugessometimes contain somewhat more spe-cies than an equivalent area in one largerefuge. This argument is scarcely rele-vant, since species must be weighted,not just counted; the question is notwhich refuge system contains more totalspecies, but which contains more speciesthat would be doomed to extinction inthe absence of refuges. A refuge systemthat contained many species like starlingand house rat while losing only a fewspecies like ivory-billed woodpecker andtimber wolf would be a disaster.

2) "For 'fugitive species' adding up toa small fraction of a regional biota asingle large refuge could be exactly thewrong strategy" (7). This argument isalso usually irrelevant, since fugitive spe-cies of high dispersal ability will oftensurvive well in the absence of any ref-uges.

3) Catastrophes like fire or diseasecould affect populations in the whole of alarge refuge but might not reach some ofa network of small refuges. This argu-ment is valid.

4) Implicit in a comment by Sim-berloff and Abele (their sentence "Morerealistically, we would hypothesize. . .") is the recognition that each smallrefuge might save a different member ofa set of mutually exclusive competitors,of which one would come to exclude theothers from a single large reserve. Thisargument is also valid.

Against the two valid arguments formultiple refuges must be set the clearmessage of the island dilemma: differentspecies have different minimum area re-quirements, while cases of maximum arealimits are extremely rare, and the spe-cies most in need of refuges are doomedin a system of small refuges. The ex-tinctions in the New Zealand forest re-serves and on Barro Colorado warn ushow rapidly the ecosystems of under-sized reserves can collapse to an inevi-table final solution. If the best solution ofa system of multiple large refuges cannot

SCIENCE, VOL. 193

on January 24, 2020

http://science.sciencemag.org/

Dow

nloaded from

be achieved, the best compromise wouldbe one refuge as large as possible plussome smaller refuges. This recommenda-tion is not based on idiosyncratic taxabut on a variety of taxa on at least fourcontinents. Nor is this recommendationpremature, in view of the clear message

and the rapid pace of human destructionof natural habitats. In the absence ofinput from biologists, developers may

often prefer small refuges as being easierto create and as leaving more land fordevelopment goals of obvious politicalsignificance. Biologists should familiar-ize themselves with the island dilemmaso that their arguments for large refugeswill be explicit and persuasive.

JARED M. DIAMONDDepartment ofPhysiology, UniversityofCalifornia Medical Center,Los Angeles 90024

References

1. F. W. Preston, Ecology 43, 410 (1962); R. M.May, Nature (London) 254, 177 (1975).

2. E. 0. Willis, Ecol. Monogr. 44, 153 (1974); J.Terborgh, BioScience 24, 715 (1974); in TropicalEcological Systems, F. Golley and E. Medina,Eds. (Springer-Verlag, New York, 1975); E. 0.

Wilson and E. 0. Willis, in Ecology and Evolu-tion of Communities, M. L. Cody and J. M.Diamond, Eds. (Harvard Univ. Press, Cam-bridge, Mass., 1975).

3. J. M. Diamond, in Ecology and Evolution ofCommunities, M. L. Cody and J. M. Diamond,Eds. (Harvard Univ. Press, Cambridge, Mass.,1975).

4. _, Proc. Natl. Acad. Sci. U.S.A. 69,3199 (1972); Biol. Conserv. 7, 129 (1975); inProceedings of the 16th Ornithology Congress,1974, H. J. Frith and J. L. Calaby, Eds. (Aus-tralian Academy of Science, Canberra, in press).

5. .., in Rehabilitation of Severely DamagedLand and Freshwater Ecosystems in TemperateZones, M. W. Holdgate and M. J. Woodman,Eds. (Plenum, London, in press).

6. C. A. Fleming,For. Bird, No. 196(1975),p. 15.7. D. S. Simberloff and L. G. Abele, Science 191,

285 (1976).8. T. Schoener, personal communication.9. E. L. Goldstein, Evolution 29, 750 (1976); A. E.

Galli, C. F. Leck, R. T. T. Forman, Auk 93, 356(1976); D. H. Morse, Ecology 52, 216 (1971); N.W. Moore and M. D. Hooper, Biol. Conserv. 8,239 (1975).

10. H. L. Jones, thesis, University of California atLos Angeles (1975).

11. A. R. Main and M. Yadav, Biol. Conserv. 3, 123(1971); J. Hope, Proc. R. Soc. Victoria 85, 163(1973); J. Brown,Am. Nat. 105, 467 (1971).

1 March 1976; revised 8 June 1976

Certain interpretations in the report bySimberloff and Abele (1), if accepted un-

critically, could be detrimental to effortsto protect endangered wildlife.

Following a now well-established prac-

tice, Simberloff and Abele use resultsfrom island biogeography to draw infer-ences about the efficacy of isolated parksand refuges as reservoirs of natural diver-sity. However, some of their conclusionsare contrary to those espoused by otherswho have considered the same problem(24).

Simberloff and Abele consider an ex-

periment in which very small (<0.05 ha),isolated copses of mangroves were dis-sected into several lesser units separated

10 SEPTEMBER 1976

by channels wide enough to reduce thedispersal of some of the arthropod spe-cies present. After an equilibration peri-od of 3 years, a census was again madeof the "archipelago" for its arthropodfauna, with the result that the collectionof separate islets contained a few morespecies (81 as compared to 77) than didthe original intact copse. From this theyconclude that "the more (and smaller)refuges posited as an altcrnative to asingle large one, the more likely is thearchipelago of small refuges to containmore species."A key feature of their experiment was

the presence nearby of large continuousstands of mangroves containing a rich"'source" fauna of hundreds of speciesof arthropods, many of which were ca-pable of invading tiny outlying islets.Indeed, in discussing the experiment,they recognize that "possibly the in-creased extinction rates on the individualislands in this mangrove archipelago aremore than compensated for by the pres-ence of the other islands as nearby sourcesgenerating high propagule . . . inva-sion rates." In contrast, those of uswho have argued the essentiality of largepreserves have imagined quite a differentscenario, one in which most of the land-scape has been preempted by agricultur-al or other human uses, and in whichscattered parks remain as the onlyredoubts for species that are unable toadapt to degraded habitats. The islandsconsidered by Simberloff and Abelewere at equilibrium, meaning that extinc-tions were in balance with recurrent im-migrations from a rich external source.However, the dynamics of equilibrialsystems are simply not germane to theproblem of isolated parks set in an in-tensively exploited landscape; rather,the appropriate context is that of land-bridge islands in which the source hasbeen removed and only islands remain.Under these circumstances, logic callsfor a strategy of minimizing extinctions,and this, I contend, is best accomplishedwith large preserves. As I shall explainbelow, there are circumstances in whichlarge preserves are neither necessary norappropriate, but these are special casesdirected toward particular species, rath-er th;an toward whole ecosystems.

If it is agreed that the primary objec-tive of a rational conservation policyshould be to preserve viable populationsof as many as possible of the species thatinhabited the pristine landscape, then atleast some large reserves are a necessity.This is immediately evident from the factthat species at the top of the trophicladder (such as wolves, bears, eagles,and mountain lions) require extensive

foraging ranges. Population densities ofsuch species are low, typically on theorder of one individual per 10 km2. Toprotect representative samples of com-plete ecosystems, areas of hundreds orthousands of square kilometers are es-sential.An optimal system of preserves should

be designed to minimize extinctions, amatter that Simberloff and Abele passover lightly. Much of our knowledge ofextinction rates comes from the study ofland-bridge islands, islands that were cutoff from the adjacent mainland by risingwater levels at some known time in thepast (frequently the end of the Pleisto-cene). Kinetic analysis begins with theassumption that land-bridge islands ini-tially contained a species complementequal to that of an equivalent-sized seg-ment of mainland. When dispersal is shutoff or severely restricted by the inter-position of a water barrier, the high-diversity ecosystem of the newly createdisland begins to "relax," and eventuallyconverges toward the low-diversity con-dition of a strictly oceanic island ofequivalent size, climate, and remote-ness. Several studies of land-bridge is-lands have been completed, and the re-sults are gratifyingly concordant (3-5).The following conclusions appear to bewell substantiated.

1) Species loss is area dependent. Anisland of 250 km2 is estimated to loseabout 4 percent of its resident bird spe-cies during the first century, while one of5000 km2 loses only 0.5 percent (4).

2) Extinctions proceed rapidly at firstas the most vulnerable species drop out,and then at a diminishing pace as thecommunity approaches equilibrium.

3) Among the first species to expire arethose on the highest rungs of the trophicladder, and the largest members of feed-ing guilds. The implications of this areuncertain, because the effects of toppredators, or even herbivores, on theinteractions of species in the lower tro-phic levels of terrestrial ecosystems arepoorly understood. In some aquatic eco-systems, however, it is known that theremoval of "keystone" predators canlead to dramatically altered, usually lessdiverse communities (6).

4) Where it has been possible to exam-ine replicated groups of land-bridge is-lands, the evidence suggests that the or-der of extinctions is highly consistent.One can infer from this that the individ-ual units of a scattered park systemwould lose very similar sets of species.

5) As relaxation goes to completion,the character of land-bridge island aviancommunities is gradually transformedfrom one typical of the dominant vegeta-

1029

on January 24, 2020

http://science.sciencemag.org/

Dow

nloaded from

tion (such as primary rain torest) to onetypical of successional vegetation (eventhough the quality of the habitat appar-ently remains unaltered). In other words,the end point of relaxation is a commu-nity composed largely or entirely of wide-spread "weedy" species which are ofnegligible interest to conservationists.

In conclusion, I can affirm that exten-sive areas are needed to preserve exam-ples of intact ecosystems and to forestallthe extinction of species having largespace requirements. This is not to say,however, that small refuges may not beadequate to serve more limited pur-poses, such as protecting the habitat oflocalized endemic forms or the nestingsites of colonial species. In enacting acomprehensive conservation policy wewill have to be sensitive to the distinctivefeatures of particular species and ecosys-tems and intensify the search for ways ofprotecting wildlife that do not conflictexcessively with society's needs forspace and resources.

JOHN TERBORGHDepartment ofBiology,Princeton University,Princeton, New Jersey 08540

References1. D. S. Simberloff and L. G. Abele, Science 191,

285 (1976).2. J. M. Diamond, Biol. Conserv. 7, 129 (1975); A.

E. Galli, C. F. Leck, R. T. T. Forman, Auk 93,356 (1976); R. M. May, Nature (London) 254,177 (1975); E. 0. Wilson and E. 0. Willis, inEcology and Evolution of Communities, M. L.Cody and J. M. Diamond, Eds. (Harvard Univ.Press, Cambridge, Mass., 1975).

3. J. M. Diamond, Proc. Natl. Acad. Sci. U.S.A.69, 3199 (1972); E. 0. Willis, Ecol. Monogr. 44,153 (1974).

4. J. Terborgh, BioScience 24, 715 (1974); inTropical Ecological Systems, F. Golley andE. Medina, Eds. (Springer-Verlag, New York,1975).

5. J. H. Brown, Am. Nat. 105, 467 (1971); D. A.Hooijer, Neth. J. Zool. 25, 46 (1975).

6. J. A. Estes and J. F. Palmisano, Science 185,1058 (1974); R. T. Paine, Am. Nat. 100, 65(1966); Oecologia 15, 93 (1974); __ and R.L. Vadas, Limnol. Oceanogr. 14, 710 (1969); T.M. Zaret and R. T. Paine, Scienc e 182, 449(1973).

23 February 1976

Simberloff and Abele (1) have ques-tioned recent recommendations (2) thatfaunal preservation can best be accom-plished by establishing single large re-serves. Instead, they conclude that theo-retical considerations could in certaincircumstances support the sequesteringof a series of smaller reserves, and theypresent experimental evidence in sup-port of this view. However, their propo-sition depends on biologically unrealisticassumptions and should not be applied toany practical problem of conservationwithout explicit proof that the assump-tions are true.

Simberloff and Abele address them-selves mainly to predictions arising fromclassic mathematical models of the spe-cies-area relationship (3). According tosuch models, the equilibrial species num-

1030

ber On a large island is higher than tnenumber on an ecologically similar butsmaller island, because the immigrationrate of new species is lower and theextinction rate of resident species is high-er on smaller islands. In addition, be-cause immigration rates decrease withdistance, classical models also predictthat more species should be maintainedon islands close to continental sourceareas. Simberloff and Abele accept thesepredictions but present computationsthat show that a given refuge area maysupport more species if it consists of anumber of smaller units. To obtain thisresult, they propose a model in which allof the species in the pool have equaldispersal and survival abilities or inwhich strong interactions between spe-cies make it likely that different sets ofspecies will survive on different islands,or both. However, there is no evidencethat such extreme conditions are in factapproached in complex communities (4).For example, different reptiles, birds,and mammals require different minimuminsular areas for long-term survival (5);families of birds differ in their species-area relationships (6); extinction rates onland-bridge islands differ among taxa (7);forest-interior bird species, especiallythose nesting on or near the ground, tendto disappear most rapidly from small for-ested areas in both tropical (8) and tem-perate (9) latitudes; and even mangrovearthropods differ in their ability to estab-lish and maintain populations on smallislands (10). The diversity of species sen-sitive to effects of area and isolationdemonstrates that, contrary to the sug-gestion of Simberloff and Abele, suchsensitivity is not a taxal idiosyncrasy.The inevitable consequence of such di-versity in the context of the proposedseries of small preserves would be thepreferential extinction of the more sensi-tive species and the emergence on eachisland of a species assemblage whosemembers can utilize disturbed habitatsand therefore are destined to surviveeven in the absence of preserves.A second inadequacy of individual

small faunal preserves, or series of suchpreserves, is their failure to provide areasonable facsimile of the entire func-tioning ecological community they areintended to represent. In fact, Simberloffand Abele agree with Sullivan and Shaf-fer (11) that the preservation of entirecommunities rather than single endan-gered species is a highly desirable goal.If a functioning community is to be pre-served, it must be acknowledged thatsome species, particularly large ones andthose at higher trophic levels, requireextensive areas of continuous habitat forsurvival. For example, no one would

propose that an entire mangrove commu-nity, including vertebrates and larger in-vetebrates, could be preserved in re-serves the size of the intact islands (0.02to 0.05 ha) studied by Simberloff andAbele (1), much less in the tinier archi-pelagos created, even if many islets wereinvolved.Human impact is often a serious practi-

cal problem in natural areas because rec-reational activity is usually programmedinto the reserve from the outset, or if notprogrammed, is difficult to prevent. Theeffects of such activity can be severeeven in gigantic national parks (12). Insmall reserves, human influences can bedisastrous.,

Island area, isolation, and human dis-turbance may interact in complex fash-ion. Although Simberloff and Abele referto a critique by Lynch and Johnson (13)of reported high avian turnover rates onislands, one of the main points of thecritique was that many insular extinc-tions and colonizations have been re-lated to human disturbance. However,the perturbations that result from habitatalteration and destruction, introductionof nonnative species, use of pesticides,or other stresses are becoming increas-ingly important determinants of speciescomposition and turnover. These influ-ences will intensify in the future and willhave greatest effect on small preserves.Therefore, the most prudent preserva-tion strategies are those that insulate sen-sitive species from the effects of humandisturbance by setting aside large contin-uous natural areas.

Classical theory (3) predicts high turn-over rates on small isolated islands. Toemphasize the reality of such turnoveron mainland "habitat islands" we citeresults of Kendeigh's annual censuses ofbreeding birds (14, 15) for TreleaseWoods, a 22-ha deciduous forest pre-serve in Illinois that is surrounded byagricultural land. Several patterns areevident. (i) Annual turnover, computedby the method of Diamond (16), was high(mean = 13.6 percent; range = 5.3 to27.3 percent for the years 1934-1975). Of62 breeding bird species, only nine (17)have been present in each of the 48 cen-suses since 1927. (ii) Three forest interiorspecialists characteristic of the Easterndeciduous forest (14) have not bred inthe woods during the census period; sixothers (18) have bred only sporadicallyfor a year or two at a time. It is reason-able to assume that some or all of thesespecies nested regularly (19) in the pri-mordial forest prior to European settle-ment, but had been extirpated by thetime of the first census in the relict wood-land. (iii) Ecologically generalized("weedy") species, many of them per-

SCIENCE, VOL. 193

on January 24, 2020

http://science.sciencemag.org/

Dow

nloaded from

grants, now dominate the TreleaseWoods avifauna. (iv) Between 1934 and1953, the mean species richness of theavifauna in Trelease Woods was 23.3(standard deviation = 4.9). By about1953 the third most abundant tree, Amer-ican elm (Ulmus americana), had begunto die from Dutch Elm disease. Theyears 1954 through 1975 then witnessed ahighly significant (P < .01) increase inavian species richness to an average of32.5 (standard deviation = 2.8), but pre-dictably (20), the birds that profited wereforest-edge species. Thus, an additionalinadequacy of small preserves is theirsensitivity to destabilization.Many of the edge species that colo-

nized the interior of Trelease Woods inspectacular fashion during its destabiliza-tion also colonize the edges of artificialswaths cut through forest (21). In suchcases, the overall avian species richnessin an area is increased, but only at theexpense of forest-interior species. Wetherefore hope that the experimental de-sign (4) of Simberloff and Abele, inwhich such ecotonal effects may havebeen absent, will not be used to justifydisastrous fragmentation of existing for-est by roads, pipelines, or similar proj-ects. Our objections to the creation oflong strips of edge within a single pre-serve apply even more strongly to seriesof small forest preserves. In areas whereforest is reduced to isolated woodlots,avian brood parasites, egg predators,and nonnative nest-hole competitors areusually abundant (22) in the surroundingagricultural and urban environments andoften invade small tracts. Even actingsingly, such species can exert intolerablepressure on other bird species (23), andtheir combined impact may be a majorforce in the avifaunal changes that suc-ceed forest fragmentation.We now wish to apply our general

reasoning to preservation of the Easterndeciduous forest, a system with whichwe have considerable firsthand experi-ence. Simberloff and Abele cite Ter-borgh's statement (2) that reduction andfragmentation of the forests of EasternNorth America caused the extinction of,at most, two bird species. Although it istrue that few North American birds havebecome extinct on a continental scalewithin historic times, we contend thatoptimism about the avifauna of the East-ern forest is warranted only as long asextensive areas of homogeneous forestremain standing. In fact, Terborgh point-ed out correctly (2) that important fac-tors in avian species survival were (i)the ability of most forest birds to utilizemiddle successional second growth and(ii) the retention of a total forested area,

10 SEPTEMBER 1976

much below half of the original area.Today the Eastern forest is an archi-pelago of second growth woodland frag-ments that vary greatly in size, and it istherefore possible to determine whethersubsets of these fragments are in factacting as preserves on a contemporarybasis. Breeding bird censuses and sur-veys in Eastern North America (24, 25)show that avifaunal composition of for-est fragments depends on their size. Inextensive forest tracts, up to 92 percentof the breeding individuals are neo-tropical migrants (26). However, manyneotropical migrant species disappearfrom small isolated forest tracts such asTrelease Woods, and the avifauna ofsuch tracts tend to be dominated by spe-cies that are either permanent residentsor short distance migrants (9, 25). Exten-sive census and survey work (27) in cen-tral Maryland shows that small (less than22 ha) tracts have a depleted avifaunalcomposition also characteristic of sub-urbs and parks (9), but that deteriora-tion has not occurred in fragments ofsimilar size and vegetational composi-tion adjacent to extensive forest (28).Also, countywide mapping projects (29)have demonstrated that some neo-tropical migrants apparently no longerbreed in agricultural regions where forestfragmentation is most severe. Loss orreduction in breeding densities of neo-tropical migrant individuals has occurredin a relatively undisturbed mesic forestfragment that is hundreds of years old(20) and in large urban parks of secondgrowth forest (27, 30). For example, in aforest plot within Rock Creek Park in theDistrict of Columbia, the percentage ofneotropic migrant breeders declinedfrom 87 percent in 1948 to 35 percent in1974 (30). The rapidity of decay of theoriginal avifauna in urban parks, in con-trast to slower decays (20) in relativelyundisturbed forest fragments and lack ofperceptible decay in nearby extensivehomogeneous forest (28), implicates hu-man perturbation as an important factorcontributing to the deterioration. Thusthe available data for the Eastern forest,far from demonstrating that large faunalpreserves are unnecessary, describe atroubled system in which local and re-gional extinctions of forest interior spe-cialists are commonplace and in whichlarge series of existing small, isolatedforest areas have failed to preserve, evenin contemporary time, many of the smallavian species that once dominated theforest.The final argument against fragmenta-

tion of our rapidly disappearing largeareas of relatively undisturbed habitat isthe unhappy fact that the process is, for

CLI fr1 C a, , C jI -, 1 %. . C.D£551

berloff and Abele's concern about the"cost and irreversibility of large-scaleconservation programs" (1) ignores thefact that it is much easier to convert anatural area into a housing developmentthan vice versa. Therefore, the most pru-dent strategy is to maximize reservesize. If, as Simberloff and Abele pro-pose, an alternate strategy proves moreuseful in specific instances, we anticipateno shortage of economic interests willingto fragment the preserves at a later date.We feel some responsibility to suggest

orders of size that are relevant to thedesign of preserves. Optimal size for pre-serves varies with geography and thekinds of communities involved, but thehistory of Barro Colorado Island in thePanama Canal Zone is a stern reminderof the irreversible losses that might oc-cur if the size of a preserve is in-adequate. In this instance (8, 31) an areaof nearly 1500 ha was insufficient to re-tain the characteristic avifauna of thelarger tropical forest from which the is-land was separated by canal constructionin 1914. Our analysis of the TreleaseWoods data shows that 22 ha is hopeless-ly small, even for preservation of smallforest-interior birds. We agree with theprinciple suggested by Sullivan and Shaf-fer (11) that primary reserves should beof sufficient size to support stable popu-lations of large mammals, and with Ter-borgh in his estimate that thousands ofsquare kilometers may be required toreduce extinction rates to acceptable lev-els. If such sizes are involved, thereseems to be no need for controversyabout the optimal size of forest pre-serves, since we are unaware of anyplans or opportunities to sequester areasthat would be inappropriately large.An acknowledgment of the need for

large preserves should not be miscon-strued as an argument against smallerones. Certainly, small reserves are betterthan none and can accomplish such pur-poses as (i) preservation of taxa that cansurvive in small areas, (ii) preservationof unique microhabitats, (iii) provision of"stepping stones" between larger re-serves, and (iv) provision of local educa-tional and recreational benefits. We do,therefore, encourage the sequestering ofsmall reserves whenever the establish-ment of a large reserve is not possible.

In summary, we urge that the size ofecological preserves be maximized be-cause (i) large areas have high immigra-tion rates and low extinction rates; (ii)some taxa require very large areas forsurvival; (iii) preservation of entire eco-logical communities, with all trophic lev-els represented, requires large areas; (iv)large preserves are better buffered

1031

on January 24, 2020

http://science.sciencemag.org/

Dow

nloaded from

against human perturbation and naturaldisaster; (v) large areas are necessary tominimize the pressures of predation,parasitism, and competition exerted byspecies abundant in the disturbed areassurrounding the reserves; (vi) failures ofsmall reserves, originally considered tobe adequate, have been amply docu-mented; and (vii) the irreversibility offragmentation demands a conservativepreservation strategy.

Simberloff and Abele have performeda useful service by focusing attention onthe potential pitfalls of an oversimplifiedmodel. We contend, however, that exist-ing theory corroborated by empiricaldata is sufficient to validate the generalconclusion that refuges should contain aslarge a contiguous area as possible.

ROBERT F. WHITCOMBPlant Protection Institlute,Agricultuiral Research Service,U.S. Department ofAgriculture,Beltsville, Maryland 20705

JAMES F. LYNCHChesapeake Bay Centerfor Environmental Studies,Smithsonian Institution,Edgewater, Maryland 21037

PAUL A. OPLEROffice ofEndangered Species,U.S. Fish and Wildlife Service,Washington, D.C. 20240

CHANDLER S. ROBBINSMigratory Bird and Habitat ResearchLaboratory, U.S. Fish and WildlifeService, Lautrel, Maryland 20811

References and Notes

1. D. S. Simberloff and L. G. Abele, Science 191,285 (1976).

2. J. Terborgh, in Tropical Ecological Systems, F.Golley and E. Medina, Eds. (Springer-Verlag,New York, 1975); E. 0. Wilson and E. 0.Willis, in Ecology and Evolutiotn of Commu-nities, M. L. Cody and J. M. Diamond, Eds.(Harvard Univ. Press, Cambridge, Mass., 1975);J. M. Diamond,Biol. Conserv. 7, 129(1975).

3. F. W. Preston, Ecology 43, 410 (1962); R. H.MacArthur and E. 0. Wilson, Evolution 17, 373(1963); R. H. MacArthur and E. 0. Wilson, TheTheory ofIsland Biogeography (Princeton Univ.Press, Princeton, N.J., 1967).

4. Simberloff and Abele report that one of twoartificial archipelagos constructed by channelinga small mangrove island contained marginallymore species of arboreal arthropods than thesingle island from which it had been created (81as compared to 77 species), but this small differ-ence is unconvincing.

5. J. H. Brown,Am. Nat. 105, 467 (1971); H. Heat-wole, Atoll Res. Bull., No. 180 (1975); N. W.Moore and M. D. Hooper, Biol. Conserv. 8, 239(1975); A. E. Galli, C. F. Leck, R. T. T. For-man,Auk 93, 356(1976).

6. J. Terborgh, Evolution 27, 338 (1973).7. J. M. Diamond, Proc. Natl. Acad. Sci. U.S.A.

69, 3199(1972).8. E. 0. Willis, Ecol. Monogr. 44, 153 (1974).9. R. F. Whitcomb, C. S. Robbins, D. Bystrak, M.

K. Klimkiewicz, B. L. Whitcomb, J. F. Lynch,in preparation.

10. E. 0. Wilson and D. S. Simberloff, Ecology 50,267 (1969); D. S. Simberloff and E. 0. Wilson,ibid., p. 278.

11. A. L. Sullivan and M. L. Shaffer, Science 189,13 (1975).

12. A. K. Fitzsimmons,ibid. 191,440(1976).13. J. F. Lynch and N. K. Johnson, Condor 76, 370

(1974).14. S. C. Kendeigh, Ecol. Monogr. 14, 67 (1944).

1032

15. , Audubon Field Notes 2, 232 (1948).The other 26 annual censuses are publishedin No. 6 of Audubon Field Notes, volumes3 through 29 (1949-1975). The extensive dataavailable for this plot deserve lengthy futuretreatment.

16. J. M. Diamond, Proc. Natl. Acad. Sci. U.S.A.64, 57 (1969).

17. These species were: downy woodpecker (Den-drocopos pubescens), great crested flycatcher(Myiarchus crinitus), eastern wood pewee (Con-topus virens), common crow (Corvus brachy-rhynchos), starling (Sturnus vulgaris), red-eyedvireo (Vireo olivaceus), brown-headed cowbird(Molothrus ater), indigo bunting (Passerina cy-anea).

18. Species that never bred were: black-and-whitewarbler (Mniotilta varia), worm-eating warbler(Helmitheros vermivorus), hooded warbler (Wil-sonia citrina). Species that bred sporadicallywere: Acadian flycatcher (Empidonax vires-cens), American redstart (Setophaga ruticilla),blue-gray gnatcatcher (Polioptila caerulea),ovenbird (Seiurus aurocapillus), scarlet tanager(Piranga olivacea), yellow-throated vireo (Vireoflavifrons).

19. At the time of the studies of R. R. Graber and J.W. Graber [Ill. Nat. Hist. Surv. Bull. 28, 383(1963)], all the missing species were recorded asbreeding in Illinois.

20. B. L. Whitcomb, R. F. Whitcomb, D. Bystrak,Am. Birds, in press.

21. V. H. Hebert, Audubon Field Notes 22, 687(1968); J. R. Longcore,Am. Birds 25, 997 (1971).

22. The nest predators are: blue jay (Cyanocittacristata), common crow, and common grackle(Quiscalus quiscula). Brown-headed cowbird isa brood parasite; house sparrow (Passer domes-ticus) and starling are nest-hole competitors.

23. H. Mayfield, in The Kirtland's Warbler, Bulletin40 (Cranbrook Institute of Science, BloomfieldHills, Mich., 1960), p. 144.

24. R. R. Bond, Ecol. Monogr. 27, 351 (1957).25. Censuses from large, undisturbed foresL from

Audubon Field Notes (now Am. Birds): for In-diana, ibid. 28, 1007 (1974); for Maryland, ibid.29, 1089 (1975); for Alabama, ibid. 3, 267 (1949);for Georgia, ibid. 17, 502 (1963); for Tennessee,ibid. 26, 951 (1972); for West Virginia, ibid. 29,1086 (1975); for Pennsylvania, ibid. 26, 958(1972). Censuses from small forest tracts: forGeorgia, ibid. 7, 343 (1953); for Michigan, ibid.28, 1006 (1974); for Ohio, ibid. 28, 1009 (1974).See also more extensive data from Delaware: J.T. Linehan, R. E. Jones, J. R. Longeore,Audubon Field Notes 21, 641 (1967).

26. R. H. MacArthur, Auk 76, 318 (1959); see also(25, data for West Virginia).

27. The data base for the central Maryland-Wash-ington, D.C., area, developed by many workersover three decades, includes 155 breeding birdcensuses of 49 forest plots, 93 point surveys indeciduous forest fragments, three countywidebird atlas mapping projects, and 111 breedingbird surveys of 55 routes.

28. L. MacClintock, R. F. Whitcomb, B. L. Whit-comb, Am. Birds, in press.

29. M. K. Klimkiewicz and C. S. Robbins, ActaOrnithol. (Warsaw) 14, 446 (1974).

30. J. H. Criswell,Atl. Nat. 30, 175 (1975).31. J. Terborgh, BioScience 24, 715 (1974).16 March 1976; revised 7 June 1976

We regret being cast as the betesnoires of conservation, since our report(1) was designed to strengthen con-servation efforts by eliminating relianceon a species-area equation which alonedoes not support either of two contrast-ing refuge strategies, and by tailoringconservation efforts to the idiosyncrasiesof the taxa in question. We do not agree

with Diamond (2) and Terborgh (3) thattheir data are adequate to support thehypothesis of high extinction rates forbirds on islands. With one exception, the"evidence" from land-bridge islandsrests not on observation of which specieswere originally present, but rather oninference from the present source faunaand the species-area equation. Evenwere habitat differences well quantified,which they are not, the wide variance infitting data to the standard species-areacurve (4) would make such a deductiveleap suspect. For Barro Colorado Islandat least, the original birds are docu-mented, but the island has undergonemajor vegetational change in the last cen-tury (5) and so can hardly be used as anexample of extinction following changein the single variable of area. Perhapswith long-lived animals few appropriatedata exist, but this suggests great cautionin erecting general theories about extinc-tion.With respect to the "extreme model,"

referred to by Whitcomb et al. (6), we didpoint out that this would be an "over-simplification," and then cited many ofthe same references which Whitcomb etal. use, to exactly the same end: to indi-cate how the model might be made morerealistic. We did not "pass over lightly"(3) extinctions; our third paragraph fromthe end addressed exactly this problem.Our conclusion still stands: the spe-

cies-area relationship of island biogeogra-phy is neutral on the matter of whetherone large or several small refuges wouldbe better. We repeat our earlier state-ment: "This is not a plea, then, for aspecific conservation regime, but ratherfor more comprehensive autecologicalconsideration."

DANIEL S. SIMBERLOFFLAWRENCE G. ABELE

Department ofBiological Science,Florida State University,Tallahassee 32306

References and Notes

1. D. S. Simberloff and L. G. Abele, Science 191,285 (1976).

2. J. Diamond, ibid. 193, 1027 (1976).3. J. Terborgh, ibid., p. 1029.4. P. H. Haas,Am. Nat. 109, 371 (1975).5. E. 0. Willis, Ecol. Monogr. 44, 153 (1974).6. R. F. Whitcomb, J. F. Lynch, P. A. Opler, C.

S. Robbins, Science 193, 1030(1976).19 July 1976

Juvenile Hormone and Pest Control

The search for novel chemical pestcontrol agents with improved character-istics, such as increased selectivity fortarget pests as compared to beneficial ar-

thropods and vertebrates, is a long and

difficult process for which the short-termeconomic incentives are far more elusivethan for the development of broad-spec-trum pesticides.The report by McNeil (1) appears to

SCIENCE, VOL. 193

on January 24, 2020

http://science.sciencemag.org/

Dow

nloaded from

Island Biogeography and Conservation: Strategy and LimitationsJARED M. DIAMOND

DOI: 10.1126/science.193.4257.1027 (4257), 1027-1029.193Science 

ARTICLE TOOLS http://science.sciencemag.org/content/193/4257/1027

REFERENCES

http://science.sciencemag.org/content/193/4257/1027#BIBLThis article cites 14 articles, 2 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

trademark of AAAS. is a registeredScienceAdvancement of Science, 1200 New York Avenue NW, Washington, DC 20005. The title

(print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for theScience

1976 by the American Association for the Advancement of Science

on January 24, 2020

http://science.sciencemag.org/

Dow

nloaded from