An Estimate of the Potential Evolutionary Increase in Species Density in the Polynesian Ant Fauna Author(S): Edward O

An Estimate of the Potential Evolutionary Increase in Species Density in the Polynesian Ant Fauna Author(s): Edward O. Wilson and Robert W. Taylor Reviewed work(s): Source: Evolution, Vol. 21, No. 1 (Mar., 1967), pp. 1-10 Published by: Society for the Study of Evolution Stable URL: http://www.jstor.org/stable/2406735 . Accessed: 18/12/2012 14:10

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].

Society for the Study of Evolution is collaborating with JSTOR to digitize, preserve and extend access to Evolution.

http://www.jstor.org

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions EVOLUTION INTERNATIONAL JOURNAL OF ORGANIC EVOLUTION PUBLISHED BY THE SOCIETY FOR THE STUDY OF EVOLUTION

Vol. 21 March, 1967 No. 1

AN ESTIMATE OF THE POTENTIAL EVOLUTIONARY INCREASE IN SPECIES DENSITY IN THE POLYNESIAN ANT FAUNA'

EDWARD 0. WILSON AND ROBERT W. TAYLOR2 Biological Laboratories,Harvard University,Cambridge, Massachusetts

Received September20, 1965

An exceptionallyhigh proportion of the to recognizethat the Polynesianants offer ant speciesof Polynesiaare "tramps,"that an unusual researchopportunity; namely is,species carried about the world by human thechance to comparethe synthetic, poorly commerce.A trulynative fauna is present coadapted faunulae of the eastern and as well on theislands of westernPolynesia. northernarchipelagoes with the older,na- In the easternand northernarchipelagoes, tive,and bettercoadapted faunulae of the however,there may be in fact no truly westernarchipelagoes. A detailedaccount nativespecies at all, the faunulaehere con- of the taxonomyand distributionof indi- sistingof syntheticaggregations of tramp vidual species is being given elsewhere species. Some of these species originated (Wilsonand Taylor, 1967). In thepresent ultimatelyin theIndo-Australian area, some articlewe will describecertain particular in Africa,and some in the New World phenomenawhich might be of generalin- tropics.Most have been carried accidentally terest. to the Pacific islands on European ships, certainlywithin the last 400 years and RELATIONSHIPS OF THE POLYNESIAN probablyno earlierthan the time of Cook's FAUNA firstvoyage (1768-1771). This tramp Our surveyhas includedall of theislands fauna,despite its heterogeneityand youth, fromthe Ellice group,Rotuma, Samoa, and has achievedan orderlinessof sorts.In par- Tonga,eastward to and includingthe Mar- ticular,the numberof specieson an island quesas and Easter Island, and north to is wellcorrelated with the island's area, and Hawaii. New Zealand was excludedfrom somerelated species have acquiredcomple- thecurrent analysis, since it is faunistically mentary(mosaic) inter-and intra-island verydistinct, most or all of its distributionsthat can be explainedeasily fewnative only as the outcome of competitiveex- species having been drawn directlyfrom clusion. In the course of currentstudies Australia (Brown, 1958). The Wallis on theIndo-Australian fauna we have come Islands and Futuna have been intensively collectedonly recently 1 by Mr. GeorgeHunt, Based on researchsupported by U. S. National and theirant faunulaewill be the subject Science Foundation grant no. GB1634. 2 Present address: Entomology Division, of a later separate reportby Hunt and C.S.I.R.O., Canberra, A.C.T., Australia. Wilson.

EVOLUTION 21: 1-10. March, 1967 1

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions 2 EDWARD 0. WILSON AND ROBERT W. TAYLOR

TABLE 1. A classificationof the Polynesian ant of Hygieneand TropicalMedicine Expedi- speciesaccording to origin. tion were studiedby Santschi (1928). In theperiod 1938-1962, five other entomolo- 1. Endemic to one or more Polynesian archipela- gists, 0. H. Swezey, E. C. Zimmerman, goes: Ectomomyrmex insulanus, Ponera loi, P. T. E. Woodward,G. Ettershank,and R. W. woodwardi, Strumigenys mailei, Pheidole Taylor,made additional collections. A five- aana, P. atua, Vollenhoviapacifica, V. samo- week tourby Taylor and his wifein 1962 ensis,Rogeria exsulans,Adelomyrmex samo- was devotedwholly to collectionand study anus, Camponotus navigator,C. rotumanus, C. flavolimbatus,C. conicus, C. nigrifrons, of the Samoan ants. Polyrhachisrotumana. Classificationof the 83 known Poly- 2. Continuouslydistributed from Indo-Australian nesian species accordingto biogeographic area into Polynesia: originis givenin Table 1, and thecomposi- Prionopelta kraepilini,Platythyrea parallela, tion of the faunulaeof the betterknown Ponera incerta,P. tenuis,Hypoponera con- finis,H. punctatissima,Cryptopone testacea, islands is given in Figures 1 and 2. Two Odontomnachussimillimus, Anochetus graeffei, conclusionsof immediaterelevance can be Eurhopalothrixprocera, Smithistruma dubia, drawnon the basis of thesepartitions, and Strumigenys szalayi, Pheidole fervens, P. on some additionalconsiderations of par- oceanica, P. umbonata, P. sexspinosa,Sole- ticularspecies presented elsewhere by Wil- nopsispapuana, Oligomyrmexatomus, Mono- morium talpa, Tetramoriumpacificum, T. son and Taylor (1967). tonganum, Rogeria sublevinodis,Iridomyr- First, it is apparent that, prior to the mex anceps, Tapinoma minutum, Techno- comingof man, few if any native species myrinexalbipes, Paratrechina minutula, Cam- rangedeast of Rotuma,Samoa, and Tonga. ponotus chloroticus. 3. "Tramp species": certainlydistributed by re- The evidenceis as follows: cent human commerce: to occur in Hypoponera opaciceps, H. zwaluwenburgi, No certain endemics are known Five which Trachymesopus stigma, Leptogenys maxil- Polynesia beyond theseislands. species to be peculiar to the losa, Syscia typhla, Trichoscapa membrani- prior to 1950 were considered of Pacific Ponera fera, Strumigenysgodeffroyi, S. lewisi, S. peripheral areas the [namely Syscia typhla Roger (=Cera- rogeri,Quadristruma emmae, Pheidole mega- swezeyi (Wheeler), pachys silvestriiWheeler), Quadristruma emmae cephala, Solenopsis geminata, Monomorium (Emery) (= Epitritus wheeleri Donisthorpe), destructor,M. latinode,M. floricola,M. fos- Chelaner antarcticm (White) (= Monomorium sulatum,M. minutum,M. pharaonis,Trigly- rapaense Wheeler), and Plagiolepis alluaudi Forel phothrixstriatidens, Tetramorium caespitum, (= P. mactavishi Wheeler = P. augusti Emery)] T. guineense,T. simillimum,Cardiocondyla are now known to occur elsewhere. Three others, emeryi,C. nuda, C. wroughtoni,Iridomyr- Amblyoponezwaluwenburgi (Williams) of Hawaii, mex humilis, Tapinoma melanocephalum, Smithistrumamumf ordi (Wheeler) of the Mar- Anoplolepislongipes, Plagiolepis alluaudi, P. quesas Islands, and Oligomyrmex tahitiensis exigua, Paratrechinabourbonica, P. vaga, P. Wheeler of Tahiti, are still unknown elsewhere longicornis,Brachymyrmex obscurior, Cam- but belong to poorly collected and taxonomically ponotus variegatus. little-knowngenera. Moreover, 0. tahitiensisis 4. "Tramp species" interceptedin quarantine at known solely from the sexual castes and cannot Honolulu but not yet establishedin Polynesia: even be compared with most of the other Indo- Brachyponerasolitaria, Tetramorium caespi- Australianmembers of the genus,which are known tum,Wasmannia auropunctata. only from the worker caste. Thus, the status of 5. Uncertainstatus: these threeremaining "endemics" is very dubious. Amblyoponezwaluwenburgi, Ponera swezeyi, Also, despitea plethoraof subspecificand varietal Smithistrumamumfordi, Oligomyrmex tahi- names applied in the literatureto populations of tiensis,Chelaner antarcticum. species now living in the central and eastern Pacific, we have detected only a single example of true geographicvariation in these populations. The Polynesianant fauna is now rela- The case is furthermorea rather trivial one in- tivelywell known. Samoa, the archipelago volving color and the thickness of propodeal of principalinterest to us, has been espe- spines in samples of Pheidole sexspinosa Mayr from the Marquesas and Society islands. cially well surveyed. Large collections About 20 Indo-Australianant species range to made in the 1920's by the London School some point east or north of Rotuma, Samoa, and

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions SPECIES DENSITY IN THE POLYNESIAN ANT FAUNA 3

60-

50 - PO LYNESIA: V) ALL ANT SPECIES Ui U 40 XL A C.,..) A3 *32 U- 0 30

co 2 20

Z 2 29

31 *38 01201 22 2 21 33 10 27 15 5 27 .2 251 I 2 3 25 6 343 28 b26 17.8 .7 .3 18 39 19

0.1 I 10 100 1000 10,000 AREA IN SQUARE MILES FIG. 1. Total number of ant species plottedagainst island area for each of the better collected Polynesia islands. Note the high values for the three large Samoan islands (1-3), this is due to the presence of a substantial native fauna in addition to the tramp species. SAMOA: (1) Savai'i, (2) Upolu, (3) Tutuila. TOKELAU: (4) Fakaofo. DANGER ISLANDS: (5) Motu Kotawa. SOCIETY ISLANDS: (6) Bora Bora, (7) Mehetia, (8) Huahine, (9) Moorea, (10) Raiatea, (11) Tahiti. AUSTRAL ISLANDS: (12) Rimatara, (13) Raivavae, (14) Maria Island, Northeastislet, (15) Rapa. GAMBIER ISLANDS: (16) Mangareva. MARQUESAS: (17) Eiao, (18) Hatutu, (19) Ua Pu, (20) Ua Huka, (21) Hiva Oa, (22) Tahuata, (23) Nuku Hiva, (24) Fatu Hiva, (25) Mohotane. PITCAIRN: (26). HENDERSON: (27). FLINT: (28). HAWAII: (29) Hawaii, (30) Maui,(31) Molokai, (32) Oahu, (33) Kauai, (34) Nihoa, (35) FrenchFrigate Shoals, (36) Laysan, (37) Wake, (38) Midway, (39) Kure (Ocean).

Tonga, but these might easily have been trans- fromthe Indo-Australianarea. Almostall ported there by man. For example,Iridomyrmex of theendemics have close relativesin Aus- anceps (Roger), one of the most widespread of the Indo-Australiandolichoderines, was unknown traliaor Melanesia,mostly the latter. The until recent years from east of the Solomon Is- singleexception is the Samoan Rogeriaex- lands. In 1955 it was collected on Aitutakiin the sulansWilson and Taylor,which apparently Cook Islands, and in 1956 at Nandi, the inter- belongsto a group otherwiseknown only national airport communityof Viti Levu, Fiji. fromthe NeotropicalRegion. Intensive collectinghas not yet revealed it in the intermediatelysituated Samoan islands, and the STABILIZATION OF THE NEWLY case for its transportto Aitutaki by human com- ASSEMBLED FAUNULAE merce is thereforestrong. Several others of the Indo-Australianelements in the central and east- The central and eastern archipelagoes ern Pacific are certainlyknown to be tramp spe- have thusbeen populated,in largepart or cies, having become establishedin the New World even entirely,by species carriedthere by as well. Three others-Odontomachus simillimus human commerce. Some of the Indo- Fr. Smith, Tetramorium pacificum Mayr, and Australianspecies mighthave been stow- Pheidole fervensFr. Smith-have been intercepted awayson thecanoes of theearly Polynesian at quarantinestations in Hawaii, and the last two have been taken in quarantinein New Zealand. voyagers.But others,originating ultimately in nativepopulations in Africaand theNew Second, the native species of western World, must have come with European Polynesia are drawn almost exclusively ships no earlierthan 400 years ago. The

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions 4 EDWARD 0. WILSON AND ROBERT W. TAYLOR

1V) u- 40

uJ m/) POLY NES IA: 30 Oi- TRAMP SPECIES ONLY *32 0 Cf sA2 C - 20 I

D 336 29

z 0 .38 .26 *9 10 .31 33 4 ~~1216 *j 224 021 5 .28 26r 6* I? *23 *27 25 15 *8 .19 .7 37 .181 "3

0.1 I 10 100 1000 101000 AREA IN SQUARE MILES FiG. 2. Number of tramp species plotted against island area for each of the better collected Poly- nesian islands. The number code is the same as in Figure 1. With the native species removed, the Samoan islands are more consistentwith the remainderof Polynesia. questionarises then: Are the islands con- forland and freshwaterbirds in the same tinuingto fillup as quicklyas new species part of theworld (MacArthurand Wilson, reachthem, or are th'especies densities sta- 1963). Thus, numbersof species seem to bilizingthrough competitive interactions? have been limited,and this seems to have Severallines of evidencesuggest that some occurredin an orderlyfashion. degree of stabilizationhas occurred: 2) One of the moreconvincing forms of 1) Althoughthe 38 trampspecies have evidenceof competitivedisplacement is a been widelyand-judging fromHawaiian detailedmosaic complementarityof distri- and New Zealand quarantinerecords-re- butionbetween ecologically similar species. peatedlycarried through the Pacific, only a This phenomenonoccurs in severalsets of fractioninhabit any given island at one the Polynesiantramp species. The clearest time. Twenty-eightoccur on Oahu, an exampleinvolves the large,aggressive spe- island which is both relativelylarge and cies of Pheidole. P. fervensFr. Smith,a whichhas the mostactive foreigntrade of widespreadIndo-Australian element, is un- all thePacific islands and, with it, probably knownfrom Samoa at thepresent time, but the highestimmigration rate of trampant it is a dominantant in the SocietyIslands. species. Most islandshave fewerthan nine P. megacepphala(Fabricius), a pantropical species, or one-fourththe numberin the speciesof Africanorigin, well known for its available pool. As seen in Figure 2, there competitiveinteractions with other ant spe- is a good correlationbetween the area of cies,is dominanton Upolu in Samoa,but it each island and the numberof trampspe- is rare or absent in the Society Islands. cies on it. The slope of the log-logplot of P. oceanica, anotherIndo-Australian ele- the same data is approximately0.22. This ment, replaces megacephala on Savai'i, comparesfavorably with the value of 0.14 Samoa, and occurson Upolu only on the obtainedfor native ant speciesof the Solo- westernside facingSavai'i; it is relatively monIslands fromthe data of Figure4. It uncommonin the Society Islands. Else- is intermediatebetween the low (and prob- where in Polynesia the complementarity ably underestimated)value of the Solomon among the three species is maintained. Islands and the values of 0.4-0.5 obtained Fervensoccurs in Tonga and Pitcairn;it is

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions SPECIES DENSITY IN THE POLYNESIAN ANT FAUNA 5 only occasional in the Marquesas and is quite unknownin Hawaii. Megacephalais unknownfrom Tonga and Pitcairn,but it is dominanton theMarquesas and in Hawaii. Similarcomplementary patterns also occur betweenCardiocondyla emeryi Forel w4/ and C. nuda (Mayr) and, less clearly,be- Cf) tweenParatrechina bourbonica (Forel) and P. vaga (Forel). In Hawaii Solenopsis w - geminata(Fabricius) is displacedto drier a: habitats by Pheidole megacephala. On smallislands in the Dry Tortugas,Florida, z and aroundPuerto Rico, thesetwo species replaceone anotheralmost entirely, so that x both are seldomfound on the same island. w Our data are notyet sufficient to determine o /4 whetherthis is also true in Polynesia, o thoughit is stronglyimplied by records fromsmaller islands outsidethe Hawaiian group. Finally, Iridomyrmex humilis (Mayr) is proceedingto eliminateall but a fewant specieswithin its restrictedarea of distributionin Hawaii, a habitit has ex- 22 35 38 43 hibitedin otherparts of the worldwhere NUMBER OF SPECIES it has been introduced. FIG. 3. An equilibriummodel (based on Mac- Not all ant generahave developedcom- Arthurand Wilson, 1963) of the ant faunula of plementarydistributions or shown other Upolu, Samoa. An attempt is made to estimate the probable increase in the numberof species of obvious signs of competitiveinteraction. tramp originfrom the present"quasi-equilibrium" The fourspecies of Tetramorium,for exam- (at 22 species) that would resultif the extinction ple, are oftensympatric, as are the five rate were lowered,by means of evolution (E --), speciesof Ponera foundin Samoa. toward the native level. The number of known native species on Upolu is 35, and this numberis 3) Some of the trampspecies have quite taken as the minimumto which the tramp quasi- unstablepopulations, a factwhich has been equilibriumcould move if given enough time. nicelydemonstrated by E. S. Brown(1958) in his studiesof the coconutgrove species has in- of the Solomon Islands. There is some its apparentnemesis P. megacephala the evidence that fluctuations,perhaps great creased in abundanceon Upolu since 1920's. enoughin magnitudeto produceoccasional To summarize,the numbersof tramp extinction,have also occurredin Polynesia. species on individual Polynesian islands this In the earlieryears of century,Tapi- have remainedwell below 38, the number noma melanocephalum (Fabricius) was available fromthe total speciespool. The commonenough to be a house pest in stabilizationhas evidentlybeen effectedin Honolulubut apparentlyhad disappeared, part by competitivereplacement within at at leasttemporarily, by 1949 (Clagg, 1957). least twoor threegroups of relatedspecies. Trachymesopusstigma (Fabricius) was ap- Conspicuouspopulation fluctuations have parentlycommon in Samoa up to and in- occurredin some species in this century, cluding1940, but has notbeen encountered probablyresulting in a fewextinctions; the by collectorssince 1956. Pheidole fervens role of competitionin the fluctuationsis was collectedin Samoa in 1926 and 1940, suspectedbut not proven. Detailed his- but has not been foundsince 1956; while toriesof most individualspecies in Poly-

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions 6 EDWARD 0. WILSON AND ROBERT W. TAYLOR

60 45. *43

.44 50 - SOLOMONS: + n 0* ALLANT SPECIES LU 0 TRAMPSPECIES ONLY U 40-

v)U- *4 2 Q 30 * 4' z *~~~~~~~~~~~~~~40 LU

20 z

~ ~~~~~~~~~042041 44 4504 '7I L o40 01 [ I 0 100 1000 10,000 AREA IN SQUARE MILES

FIG. 4. Numbers of ant species plotted againstisland area for the betterknown of the Solomon Is- lands; based on the data of Mann (1919). The large closed circlerepresents the estimatedtotal number of species that would occur on an island with the same area as Upolu (430 square miles). The large open circle representsthe estimatednumber of thesespecies of tramp origin on the same island. (40) Florida, (41) Malapaina, (42) Ugi, (43) Malaita, (44) San Cristoval, (45) Santa Ysabel. nesiaare notyet available. The accountby Now suppose the faunulais allowed to Wilsonand Taylor (1967) is intendedas a remainat equilibriumlong enough for evo- firstchapter of a recordto be carriedon in lution to occur. In evolutionarytime the the future. speciesbecome better adapted to the local island environmentand to each other. As THE POTENTIAL INCREASE OF SPECIES thishappens, population stability should in- DENSITY THROUGH EVOLUTION crease,since those species unable to achieve The data just presentedcan be applied it will be preferentiallyeliminated. As a to the followingabstract problem. Suppose consequencethe species extinction rates will a new faunulawere assembled on an island decrease and, providingthat immigration froma pool of species chosen at random rates are not altered,species densitywill fromvarious parts of the worldand hence increase. relativelyill-adapted to each other. Im- Hence, in evolutionarytime, our imagi- migrantsare placed on the island continu- narynewly assembled fauna is notin a true ously fromthe beginningof the episode. equilibriumbut ratherin a quasi-equilib- As the numberof speciesis increased,the rium. Over a periodof a fewgenerations, rateof extinctionwill increase until in time such as mightbe observedin a humanlife- it comes to equal the rate of immigration time,the rate of speciesextinctions would (in species per year). At that point of appearto equal therate of speciesimmigra- coursethe numberof specieswill cease to tions. But followedthrough a muchlonger change. Let us assumethat the numberon periodof time,it couldbe seento have been the island reachesthe equilibriumat some almostimperceptibly less. The speciesden- value less than the numberin the species sityat quasi-equilibriumwas graduallyin- pool. creasingin thistime by meansof evolution.

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions SPECIES DENSITY IN THE POLYNESIAN ANT FAUNA 7

Presumablythe quasi-equilibriumnum- 3) Forty-threenative species are known berwould go on increasingindefinitely (but for all of Polynesia. They representthe at an ever decreasingrate) if the pool of pool of Melanesia-basedspecies that have immigrantspecies were large enoughand successfullycrossed the watergaps to Ro- theenvironment remained otherwise stable. tuma, Samoa, or Tonga by presumably Indeed,since the pool is finite,the ultimate naturalmeans. Since Upolu is centrallylo- determinantmust be environmentalstabil- cated and close to easternMelanesia, all of ity. As we expandour timescale fromthat the 43 species that occur in Polynesiaare of evolutionarytime, during which evolu- actual or presumablypotential colonists on tionis effective,to thatof geologicaltime, Upolu. The native immigrationrate curve in whichgeographic and climaticchanges forthe island is thereforedrawn to intersect are also effective,environmental stability the speciesaxis at 43 species. mustbecome decisive. To digressmomen- 4) Actually,only 35 of the 43 native tarily,it followsthat since tropicalregions species occur on Upolu. This is taken as are morestable than temperate ones, higher the minimumshort-term "equilibrium" quasi-equilibriacan be obtainedin them. numberfor native species and in Figure3 This factoralone mightaccount for most the immigrationrate curve and extinction or all of the differencein speciesdiversity rate curve for native species are drawn to betweenthe tropicaland temperatezones. intersectthe speciesaxis at 35 species. Returningto the concretecase, it would 5) Sincemost of thetramp species of the be of generalinterest to obtain some esti- world have spread throughPolynesia in mate of the amountof increasein species less than400 years,the trampimmigration numbersthat can occurthrough evolution. ratescan be takenas higherthan the native The Polynesianant fauna appears to pro- immigrationrates, and thecurves have been drawnaccordingly. The precisedifferences vide at least an entreeinto the problem. In are unknownbut are not necessaryfor the Figure3 is presentedthe quasi-equilibrium solutionof theproblem we have posed. on Upolu,Samoa, as it mightbe viewedby 6) Fromevidence already given, the ex- means of the equilibriummodel of Mac- tinctionrates of trampspecies appear to be Arthurand Wilson (1963). The following high. They are almostcertainly higher on factsand postulatesare introducedinto the theaverage than those of thenative species, model: in whichsurvival has oftenbeen prolonged 1) There are 38 speciesof tramporigin enoughto allow geographicdifferentiation known fromPolynesia (Table 1, Classes and even full speciation.The two extinc- 3-4). These constitutethe pool of avail- tion rate curves are drawn accordingly. able immigrants.If all the 38 specieswere Againthe precise differences are notneeded establishedon Upolu, the immigrationrate forthe solutionto the problem. wouldof coursebe zero because no further A fortunatecircumstance that permits specieswould be availableas colonists.We themodel to be appliedin thiscase is that have thereforedrawn the trampimmigra- the numberof species in the tramppool, tionrate curveto intersectthe speciesaxis 38, and the numberin the nativepool, 43, at 38 species. are nearly the same. The immigration 2) Actually,only 22 trampspecies are curvesof the two groupsthus descendto knownto occuron Upolu. For reasonsal- nearlythe same point on theabscissa. Also, readygiven we are postulatingthat thisis by definition,the two extinctioncurves a short-term"equilibrium" value, or nearly ascendfrom the same point, the origin. By so. In Figure3 theimmigration rate curve taking advantage of the circumstanceit and extinctionrate curve for tramp species shouldbe possibleto makepredictions con- on Upolu are thereforedrawn to intersect cerningchanges that would occur in the the speciesaxis at 22 species. equilibriumnumbers if the slopes of the

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions 8 EDWARD 0. WILSON AND ROBERT W. TAYLOR

100 (V) Lu

LLJ 101000 0

LU z 0.1 I 10 100 1Q00 10,000 AREA IN SQUARE MILES

FIG. 5. The Upolu estimateextended to islands of differingarea, at least for the Samoa region. It is assumed that the species pool remainsat 38 and that the slope of the species-areacurve does not change in evolution. (a) The presentcurve based on the relativelyhigh values for Upolu and Fakaofo. (b) and (c). inrreases of 1.5 Y and 2 Y resnectivelv. immigrationand extinctioncurves were speciesnumber be increasedthrough evolu- altered. tionarylowering of the extinctioncurves? The particularalteration of interestis This is a specificcase of the moregeneral the loweringof the trampextinction curve problem-byhow much will species density fromits presenthigh level throughevolu- increaseas evolutionoccurs in a newlyas- tion (indicatedin Figure 3 as vectorE) sembledfauna? First,it can be affirmed toward the lower, native species level. parentheticallythat the equilibriumnum- Notice thatin a newlyassembled faunula, ber, or more preciselyquasi-equilibrium in whichimperfectly adapted specieswere number,in the newly assembled tramp drawn froma much larger but constant faunula must be about 22, the number pool, immigrationrates would tend to re- actuallyoccurring there. This is a number main unchanged,since the recipientisland quite consistentwith the species-area curve (or continent)is not apt to influencethe presentedby theeastern Polynesian faunu- makeupof the sourcefauna in any signifi- lae, which are made up more purely of cantway. [In theIndo-Australian ponerine trampspecies (Fig. 2). In additionto these fauna,for example, the ratio of exchangeof trampspecies, there are on Upolu 33 native species in older islands is approximately species and one additionalspecies of un- equal to theratio of the areas,with the re- certainclassification. These species have sult that a continentor large island will stoodup to theonslaught of the trampspe- dominatea smallisland in reciprocalinflu- cies by livingin the extensivetracts of na- ence (Wilson,1965).] On the otherhand, tive vegetationextant on Upolu. Because theextinction rates in thenewly assembled the greatmajority of the same speciesalso faunulacan be expectedto be sensitiveto occuron Savai'i, a largerand evenless dis- evolution,since the rateswould tendto be turbedisland about I1 miles to the west, lowered by character displacementand we concludethat not many extinctions have otherforms of local adaptation. occurredamong the native species of Upolu The questionwe wishto pose and tryto due to competitionfrom tramp species. answeris: In the case of an island like The trampspecies are concentratedin the Upolu,by howmuch would the equilibrium cultivatedareas and the native species in

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions SPECIES DENSITY IN THE POLYNESIAN ANT FAUNA 9

thenative forests, with the two groups over- Estimatescan be made for increasein lapping widely in their habitat choices. densityfor the whole range in islandareas, Thirty-threeis the minimumquasi-equilib- at least forthe Samoa area, if it is recalled riumnumber for native Upolu ants,and we that the species-area curves vary little take it to be not far fromthe true figure. amongdifferent faunas. Such an extended We concludethat if the newlyassembled set of estimatesis givenin Figure5. fauna of tramp species were allowed to evolveon an island such as Upolu,without SUMMARY competitionfrom natives and withthe im- 1. Few if any ant speciesare nativeto migrationrate held constant,the number Polynesia east of Rotuma, Samoa, and would increasefrom about 22 to at least Tonga. The centraland easternarchipela- 33, or by a minimumfactor of about 1.5. goes have been populatedin largepart by Whatis themaximum to whichthe quasi- 35 "tramp"species carried there from vari- equilibriumwould increase over a verylong ousparts of the tropics by human commerce. time? The answerof course must be no Several othertramp species are knownto greaterthan 38, the numberin the entire have been carriedto thePacific but are not trampspecies pool, or by a factorof ap- yet establishedin the centerand east. The proximately2, unlessspecies multiplication trampspecies also occur,along with native occurred.There is another,nontrivial line ones,in westernPolynesia and outsidethe of evidencethat suggests this upper limit to Pacific area. hold even in the presenceof a muchlarger 2. No one island contains all of the pool of immigrantspecies and higherim- trampspecies, and mostcontain less than migrationrates. In the SolomonIslands, one-fourthof them. Several lines of evi- the fauna is much closerto the very rich dence suggest that the species densities sourcefauna of New Guinea and contains have stabilized. Competitivereplacement correspondinglyhigher numbers of stocks has evidentlyplayed a rolein thestabiliza- and species. Even so, theexpected number tion. of nativeant specieson an islandthe size of 3. The species numbersare considered Upolu (whosearea is 430 square miles) is not to be restingat perfectequilibrium only about 45, as shownin Figure4. but rather to be at most at a "quasi- To summarize,it is to be expectedthat equilibrium."If the immigrationrates of speciesdensity of a faunain quasi-equilib- the 38 trampspecies (35 establishedand riumwill increaseto some extentas local 3 othersavailable) wereheld constant,and evolutionoccurs. In the case of a tropical afterall the specieshad an opportunityto islandthe size and positionof Upolu,as the colonize a given island, the numberat faunulachanges specifically from a newly quasi-equilibriumcould stillbe expectedto assembled,poorly integrated condition to increase slowly as local adaptation pro- one in whichit is adapted enoughto pro- ceeded throughevolution. duce some species endemicto the island, 4. By comparingthe newlyassembled the numberof species can be expectedto trampfaunulae with older native faunulae increaseby a factorof between1.5 and 2. of Indo-Australianorigin in Samoa,an esti- If the newlyassembled faunula is better mate of the potentialincrease in species adapted,as it wouldbe if it weredrawn en- numbersthrough evolution was made. On tirelyfrom species native to a nearbyarea, an islandwith the size and historyof Upolu the lowerlimit would probably be less. In the factorof increaseshould be between any case, it is hard to imaginea situation 1.5 and 2. in whichthe species number would be more 5. Estimates can then be made for than doubled,unless the initialassemblage islands of all areas if the slope of the weredeliberately chosen for incompatibility changing species-area curve holds constant amongthe species. (Fig. 5).

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions 1CA MATARPT a WXTTTITONT ANDVT OPRl RT W7T TAVT OR

LITERATURE CITED MANN,W. M. 1918. The ants of the British BROWN, E. S. 1958. Immaturenutfall of coco- Solomon Islands. Bull. Mus. Comp. Zool. nutsin the SolomonIslands. II. Changesin Harvard63: 273-391. ant populationsand theirrelation to vegeta- SANTSCHI,F. 1928. Formicidae.Insects of Samoa tion. Bull.Entomol. Res. 50: 523-558. 5: 41-58. BROWN, W. L. 1958. A reviewof the ants of WILSON,E. 0. 1965. The challengefrom related New Zealand. Acta Hymenopterologica1: 1-50. species,p. 7-24. In Baker,H. G. and G, L. CLAGG, C. F. 1957. (Report on Exhibition.) Stebbins[ed.] The Geneticsof ColonizingSpe- Proc. HawaiianEnt. Soc. for1956, p. 197. cies. AcademicPress, New York. MACARTHUR, R. H., AND E. 0. WILSON. 1963. WILSON,E. O., ANDR. W. TAYLOR.1967. The An equilibriumtheory of insular zoogeography. ants of Polynesia. Pacific Insects Monographs Evolution17: 373-387. (in press).

This content downloaded on Tue, 18 Dec 2012 14:10:24 PM All use subject to JSTOR Terms and Conditions