Niches and Coexistence of Communities in : Repeated Patterns1

Juan A. Torres Departamentode Biologia, Colegio UniversitarioTecnol6gico de Bayam6n, Bayamon, Puerto Rico 00620, U.S.A.

ABSTRACT I studiedant coexistencein adjacentareas of upland tropicalforest, grassland, and agriculturalland in San Lorenzo,Puerto Rico. Data on food utilization,daily activity,nesting sites, microhabitatutilization and interspecificaggression were collected.' toleranceto 45?C was determinedin the laboratory. Agriculturaland grasslandants eat grain,liquid food or ,or grow fungus;only liquid drinkingand insectivorousants occurredin the forest.Some specieswithin food groups in the agriculturaland grasslandants differedin size of food consumed or daily activity.Some from agriculturalland were restrictedto differentcrops, but spatial microhabitatutilization contributedlittle to speciesseparation. Forest species differed in theiruse of litterdepths, daily activity and microhabitat.Results of interspecificaggression were influencedby priorityeffects and microclimate.I suggest that some species coexist by non- equilibriumprocesses or by stochasticevents.

RESUMEN Se estudi6 la coexistenciade hormigasen areas contiguasde bosque tropical,pradera y zonas agricolasen San Lorenzo,Puerto Rico. Se coleccion6informaci6n sobre habitos alimenticios,patrones de actividad diaria, lugaresde anidaci6n y utilizacionde microhabitats.Se estudioen el laboratoriola toleranciade las hormigasa la temperaturade 45?C. Se encontr6que las hormigasde la zona agricolay de praderacrecen hongos, consumen semillas, toman alimentoliquido o consumeninsectos. En el bosque se encontraronsolamente las hormigasinsectivoras y las que consumenalimento liquido. Dentro de estos grupos alimenticiosalgunas especiesdiferian en el tamaniode alimentoconsumido o patr6nde actividad diaria. Varias especiesde la zona agricolaestaban restringidas a diferentescosechas, pero la utilizaci6nespacial del microhabitatno contribuy6a la separaci6nde especies.Las hormigasdel bosque diferianen la utilizacionde la hojarasca,actividad diaria y microhabitat.Los resultadosde la agresi6ninterespecifica estuvieron influenciados por efectosprioritarios y microclima.Sugiero que algunasespecies coexistendebido a procesosde desequilibrioy eventosprobabilisticos.

ANTS ARE UNIQUE among insectsin theirdominance as a stabilityon the temperatezone and some oceanicislands, group (Wheeler 1910). They have veryfew natural ene- ants have served as models for the documentationof mies (Wheeler 1910, Skaife 1964, Wilson 1971), and competitiontheory in ecology(Brian 1956; Brian et al. the effectof predatorson non-reproductiveworkers is 1966; Wilson 1971; Levins et al. 1973; Culver 1974; betterdescribed as a parasiticrelationship (at the colony Lieberburget al. 1975; Davidson 1977a, b; Taylorand level). Lossesthrough foraging mortality are usuallycom- Adedoyin1978; Whitfordet al. 1981). Nonetheless,few pensatedby the amountof foodtaken in exchange(Car- studiesdealing with the factorsthat structure communi- rolland Janzen1973). Antcolonies are remarkablystable ties in the complex tropicalant faunas have been con- due to thelongevity of theirmembers, especially the queen ducted(Carroll and Janzen1973). (Wheeler 1910, Wilson 1971). Populationstability is In PuertoRico, ants are the most abundantinverte- enhancedby the accessibilityof the brood as food when brates(Levins et al. 1973). I attemptto explain how starvationthreatens; and larvae,workers and queens can resourcesare partitionedamong the speciesfor three ad- oftenendure long periodsof timewithout food or water jacentant assemblageson the island. Also, I attemptto (Brown 1973, Carrolland Janzen1973). Antsthat have answer the followingquestions: does niche dimension storageareas (granivorousants) or honeypot repletesare compensation(Schoener 1974, Fuentes 1976) exist on furnishedwith anothersafety valve againstfood short- theseassemblages; does microhabitataffect the outcome ages. of interspecificcombats among ants; do species whose Due to the low predationsuffered by ants and their nichesare includedin those of otherspecies tend to be moreaggressive; and do priorityeffects exist in theseas- semblagesthat affect the outcome of interspecificant com- I6Received 7 January1983, revised9 January1984, accepted bats? Torres(1984) explainsthe factorsthat determine 16 February1984. the numberof ant speciesin thesecommunities.

284 BIOTROPICA16(4): 284-295 1984

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions MATERIALS AND METHODS of 20 baits wereplaced in sunnymicrohabitats). Micro- habitatutilization was calculatedby countingthe number My studysite was in San Lorenzo,Puerto Rico, wherea of times that a given species was found in the baited dryseason (Decemberto April) withan averageprecipi- microhabitats. tation of 11.30 cm per month (SD = 3.35) alternates An environmentalchamber (Environator Model witha wet season (May to November)with average pre- WL221) witha glass coverwas used to determineants' cipitationof 21.74 cm per month(SD = 2.39). Average toleranceto 45?C. Ants were placed in the chamberat monthlytemperature was 22.2?C (SD = 0.65) in the dry 45?C in vials withperforated caps. The timerequired for seasonand 24.5?C (SD = 0.54) in thewet season (NOAA 50 percentof the antsto assumea crouchedposition was 1979). used as a measureof toleranceto 45'C. Controlants were FromJuly 1978 to September1980, I studiedthe kept in vials outsidethe chamber.I used a minimumof ants in threenearby areas: upland tropicalforest, grass- ten ants per vial, and repeatedthe experimentwith in- land, and agriculturalland. Two siteswere studied in the dividualsfrom ten coloniesfor most species. upland tropicalCarite forest: one near route 7740, Km A species list for these communitieswas compiled 4.4; the other near Route 184, Km 18.1. Ewel and usingbaits, berlese funnels, pitfall traps, sifting the litter, Whitmore(1973) classifiedthis area as a subtropicalwet and searchingfor nests. The taxonomyof the various forest.The tightcanopy produced a cooler and wetter monomorphicSolenopsis has not been accuratelydeter- microdimatethan in grasslandand mostagricultural land. mined(Roy Snelling,pers. comm.) and forthis reason I The ant specieswere identical in the forestedsites. will use lettersto differentiatethem. The grasslandand agriculturalareas were located near Route 7740 (Km 1.1-2.5). The grasslandwas dominat- RESULTS ed by thegrass Chlorisradiata, whichwas used forcattle grazing.The agriculturalareas studiedconsisted of plan- SPECIES AND NESTING SITES.-Table 1 presentsthe list of tain(Musa paradisiaca), coffee(Coffea arabica), and yam ant speciesfound in thesecommunities. In theapparently (Dioscoreaalata) crops.The coffeegroves resembled an appropriatenesting places, I foundthat 64 percent(N = early-successional-stageforest with a microclimatesimilar 621) were emptyin the agriculturalland, 43 percent to thatof the forest. (N = 400) in the grasslandand 82 percent(N = 500) I collecteddata on feedingby observingfour or more in the forest.This suggeststhat ants were not limitedby ant nestsof each speciesfor five minutes per hour. Food lack of nestingplaces. Competitionfor food is difficultto that ants were carryingwas removedwith forcepsor a separatefrom competition for nesting sites. The suitability piece of adhesive tape, put on a microscopeslide and of a nestingsite in the faceof competitionvaries with the examinedunder a dissectingmicroscope. The size of food amountof food the colonycan harvestfrom that partic- takenwas measuredwith calipers. Ants returningto the ular point (see Carrolland Janzen1973). nestwere checked for gastric distention which occurs when theybring liquid food.I classifiedfood in broadcategories FOOD TYPES.-Tables 2, 3, and 4 summarizethe food thatI believeare importantto the ants. Poor visibilityin data whichwere used in calculatingfood nicheoverlaps the forestthwarted most food data collection. for21 speciesin agriculturalland, 17 in grasslandand 5 To studyspatial utilization in forestlitter I took 100 in theforest (Torres 1981). The overlapmatrices for each samples,each 25 cm2,of each first,second, and third2.5 communitywere treatedas matrixof correlationcoeffi- cm layersof litter.The litterwas siftedand the ants of cientsto performa principalcomponent analysis (PCA). each species counted. At each studysite I investigated Results of the PCA were used to classifythe ants into severalapparently appropriate nesting places and counted functionalgroups. the nestsof each species.Possible nesting places included The agriculturalland antsare dividedinto four func- dead twigs,grounded logs, beneath rocks, 25 cm2ground tionalgroups (Table 1): thosethat either eat insectsand quadrats,trees, and stumps. insectparts, bring mostly liquid food in the crop (nectar, To studymicrohabitat preferences I placed tuna fish honeydew, and fruitjuices, etc.), eat grain, or baits on the ground,rocks, litter and vegetation.I put grow fungus.The fungus-growingants use similarveg- the baits at a randomlychosen time at a distanceof 5- etable matterand fecesto grow theirgardens. Although 15 m fromeach other.This was done in sunny,shady, I placed Solenopsisgeminata in the granivorousgroup, it wet, and dryareas, during night and day. Speciesfound also tendsinsects on the rootsof manygrasses. Under- in the bait and insidea 25 cm2quadrat aroundthe bait groundfeeding was not quantifiedand forthis reason S. were recorded.I sampled fourreplicates in each micro- geminatais excluded fromsome analyses. Platythyrea habitatevery month for a year.This designprovided data punctata'sextensive consumption of lepidopteranand co- on daily and seasonal patternsand was used to study leopteranlarvae separates it fromthe otherinsectivorous aggressionamong ants (exceptin the forestwhere a total ants. Ants that bringliquid food consume insectsand Ant Coexistence and Niches 285

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions Grasslandants are groupedin the same manneras in TABLE 1. Antspresent (X) in the habitats. Diet type: 1 = insectivorous;2 = liquidfood; 3 = granivorous;4 = agriculturalland (Table 1). The fungus-growingants use fungusgrowing. similarsubstrates for their gardens. Tetramorium bicari- natumand Pheidoleexigua switchto the insectivorous Agricul- group,but continueto bringgreat amounts of liquid food Grass- tural For- in theircrops. Wasmanniaauropunctata appears in the Species (code) land land est insectivorousgroup, but I believe belongsto the liquid PONERINAE food group (observationof gastricdistention is difficult Amblyoponesp. (Am.s) X in thisspecies). Comments on Monomortiumfloricola and Anochetuskempfi (An.k) X X M. ebeninumin agriculturalland observationsapply here. Anochetusmayri (An.m) X X X Hypoponeraopacior (Hy.o) X X X For the few forestspecies studied only two groups Hypoponeraergatandria (Hy.e) X X can be defined:the liquid food and insectivorousgroups Odontomachusbrunneus (Od.b) X 1 X 1 (Table 1). The fungus-growingants were absent in the Pachycondylastigma (Pacs) X X X forest. Platythyreapunctata (Plip) X 1

FORMICINAE FOOD SIZE AND MICROHABITAT.-Tables 5, 6 and 7 present Brachymyrmexheeri (Br.h) X 2 X the microhabitatutilization data, and Figure 1 the food- Brachymyrmexminutus (Br.m) X X 2 X Camponotusustus (Camu) X 1 X size distributionswhich were used in calculatingoverlap Camponotussexguttatus (Cams) X estimates(reported in Torres 1981). Data on micro- Myrmelachistaramulorum (Myrr) X X X 2 habitathave spatialand temporalcomponents. Emphasis Myrmelachistasp. (Myrs) X is given to the microspatialcomponents; the temporal Paratrechinamicrops (Pa.m) X X componentis discussedlater. Paratrechinalongicornis (Pa.l) X 2 X 2 Paratrechinasteinheili (Pa.s) X 2 X 2 X 1 NICHE DIMENSIONCOMPENSATION.-I testedthe hypothesis DOLICHODERINAE of nichedimension compensation (Schoener 1974, Fuentes Iridomyrmexmelleus (Ir.m) X X 2 X 2 Tapinomalittorale (Ta.l) X X 1976) usingfood-type, food-size and microhabitatover- Tapinomamelanocephalum (Ta.m) X X laps. The median overlapwas found on each niche di- mension.I assigneda positivesign when the overlapin Cardiocondylaemeryi (Ca.e) X 1 X 1 one dimensionfor each pair of species was above the Cardiocondylanuda (Ca.n) X 1 X 1 medianand a negativesign when below. Overlapsin the Cardiocondylavenustula (Ca.v) X X threedimensions are statisticallyindependent if theyfol- rimosus X X 4 Cyphomyrmex (Cy.r) 4 low theexpected distribution, ? (++ +), 3/8(+ + -), 38 Eurhopalothrixsp. (Eu.s) X Monomoriumebeninum (Mo.e) X 2 X 2 (+--), ?8 ( - -). Excessesand deficienciesof (+ + +) Monomoriumfloricola (Mo.f) X 2 X 2 implyredundancy and compensation,respectively. There- Mycocepurussmithi (My.s) X 4 X 4 fore,I reexaminedthe groups definedin the food-type Oxyepoecussp. ? (Ox.s) X X sectionto check forcompensation or redundancy(com- X 1 Pheidoleexigua (Ph.e) X 2 pensationor redundancycan be foundwithin subsets even Pheidolefallax (Ph.f) X 1 X 1 Pheidolemoerens (Ph.m) X X 1 if the whole does not reflectany pattern).This time I Pheidolesculptior (Ph.s) X X assigneda positivesign for all membersof a guild as Pheidolesubarmata (Phsu) X 1 X 1 definedby the food-typedimension and checkedthe sign Rogeriasp. (Ro.s) X X patternin theother dimensions. A negativesign indicates SolenopsisA (So.A) X X 2 X SolenopsisB (So.B) X X X 2 compensationin that dimension.Negative signs were SolenopsisC (So.C) X X X checkedfor induded niches (Miller 1968). A symmetrical Solenopsisgeminata (So.g) X 3 X 3 measureof overlapusually gives low valueswhen we have Strumigenysgundlachi (St.g) X includedniches, when in factthere is 100 percentoverlap X 1 X X Strumigenyslouisianae (St.r) 1 forone species. Overlap in these cases is inverselypro- Strumigenysrogeri (St.r) X X X Tetramoriumhicarinatum (Te.b) X 1 X 2 portionalto theregion of nichespace occupiedexclusively Tetramoriumcaldarium (Te.c) X X by one species. Wasmanniaauropunctata (Wa.a) X 1 X 2 For the agriculturalland species,the observedsign Wasmanniasp. (Wa.s) X distributionfollows the expected distribution for no niche compensation(G= 0.14, df = 3, P > 0.98, # species parts secondarily, except for Tetramoriumbicari- pairstested = 36). Though compensationor redundancy natum. Monomoriumebeninum and M. floricola have very in overlapswas not found when the whole assemblage similar diets and are more diversifiedthan those of the was analyzed,I checkedfor sign patternswithin guilds. other species in the liquid food group. I found that microspacedoes not separateants in the

286 Torres

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions TABLE 2. Food itemscarried by ants in agriculturalland. See Table 1 forspecies code.

o~~~~~~~~~~~~~~~~~~~~~Jt -0~~~~~ -o ,, 0S

Species E 4 oU c 8 ? c t 0

Br.m 20 20 Camu 228 3 7 13 1 1 1 3 Ca.e 154 1 20 17 3 2 4 3 4 1 7 1 5 Ca.n 112 1 11 9 2 2 4 5 Cy.r 87 3 1 1 73 1 1 3 3 Ir.m 80 29 60 9 1 1 5 4 1 2 348 16 Mo.e 92 2 66 21 1 2 9 17 1 10 21 4 88 59 Mo.f 10 7 1 6 19 3 My.s 85 15 1 1 97 7 38 5 d.b 100 2 8 2 6 1 1 4 1 1 1 Pa.e 91 16 10 24 2 1 1 1 1 119 8 Pa.s 33 2 1 1 1 2 Pl.p 132 6 2 10 1 Ph.e 34 1 5 7 1 1 3 14 Ph.f 82 7 213 30 56 3 11 6 1 2 1 8 12 2 18 4 10 Phsu 118 27 43 27 26 10 22 1 2 2 4 6 7 20 So.A 37 6 1 2 6 14 3 So.g 132102 25 14 5 3 8 1 9 9 2 7 6 St.b 95 26 3 1 1 4 Te.b 80 1 13 11 2 13 4 1 3 2 34 4 Wa.a 91 7 32 51 34 1 3 7 36 3 5 225 3 36 Total 1893 169 518 285 196 28 69 247 10 9 4 31 24 60 60 691 57 188

liquid food group nor in the insectivorousfood group. induded in T. bicarinatum'sniche. Space does not sep- Food size separatesfew insectivorous ants and fewmem- arateinsect-feeding species. bers of the liquid food group. In the case of food size, Food size separatesParatrechina longicornis from ants clear patternsof niche inclusionare observed(Fig. 1). in thegenus Monomorium in theliquid foodgroup. Mono- However,foraging space separatesthe species in theliquid moriumare includedin the nicheof P. longicornis.Space food groups fromthe insectivorousants. Insectivorous separatesthe speciesin the liquid food group fromthose ants preferredto forageon the groundand liquid food in the insectivorousgroup, but not withingroups. Insect- ants tended to forageon vegetation,suggesting redun- feedingants are foundwithin the space-nichedimension dancyin resourceutilization. A Kolmogorov-Smirnovtest of liquid food ants.A Kolmogorov-Smirnovtest shows a showsno differencein size of food used by the fungus- differencein foodsize utilizedby thefungus-growing ants growingants (D = 0.09, P > 0.05). Platythyreapunc- (D = 0.28, P = 0.05), but Cyphomyrmexrimosus in- tata, Camponotusustus, Iridomyrmex melleus and Solenop- cludes Mycocepurussmithi in its niche. sis A differfrom the otherspecies due to theirpreference Overlapsample sizes were too smallfor statistical sign forclosed-canopy crops. analysisto be performedfor the forest ants. No separation The observedsign distribution for grassland ants fol- in food size occursfor the insectivorousants. Pheidole lows the expecteddistribution for no nichecompensation moerensforages on trees,whereas Paratrechina steinheili (G= 3.42, df = 3, 0.10 < P < 0.50, # of speciespairs does not. Food size separatesIridomyrmex melleus from tested= 66). In the insectivorousants, size separates SolenopsisB in the liquid foodgroup, but SolenopsisB is Odontomacbusbrunneus from other species. Tetramorium induded in I. melleus'niche. Myrmelachistaramulorum bicarinatumdiffers from the genus Cardiocondylaand and I. melleus are similarin microhabitat(both arboreal) Pheidoleexigua in size of food used, but thesespecies are and food-sizeutilization. Camponotus ustus, a nocturnal Ant Coexistence and Niches 287

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions TABLE 3. Food itemscarried by ants in grassland.See Table 1 forspecies code.

cn~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~4

-o - C C oC o -~~~~~~~d

Species ag ; s ? >

Br.h 33 4 1 4 18 1 Ca.e 79 1 8 19 4 2 1 1 2 6 13 Ca.n 75 1 4 11 1 4 2 1 4 1 13 Cy.r 82 11 3 1 25 1 1 10 7 Mo.e 80 1 23 37 1 4 3 1 3 15 2 155 19 Mo.f 85 1 31 70 10 2 1 23 1 1 47 6 304 41 My.s 80 3 109 21 1 d.b 147 1 1 8 1 1 7 1 3 Pa.l 65 6 2 18 1 1 1 53 2 Pa.s 80 9 8 7 1 2 1 19 6 Ph.e 40 5 4 7 1 2 4 11 Ph.f 80 2 235 50 30 14 3 1 6 3 7 5 3 24 39 Phsu 80 5 12 7 10 1 1 1 1 5 1 18 So.g 81 59 1 1 2 1 9 1 1 St.b 71 14 10 1 1 1 2 2 Te.b 169 1 75 17 5 4 26 2 2 3 1 1 2 27 6 Wa.a 80 2 12 27 17 5 1 1 3 2 211 6 Total 1407 87 440 268 112 24 59 150 5 30 14 9 15 94 53 608 194

arborealant, seems to belongwith the insectivorousants; cated goodness of fit (G test) to classifyant species as food size probablyseparates it fromP. moerens. diurnal,nocturnal or both(Table 8). The expectedvalues Ant distributionswere influencedby depth in the werecalculated by dividingthe numberof baitsput dur- forestlitter (G= 588.60, df= 18, P < 0.005). I found ing an intervalof timeby thetotal number of baitsplaced threegroups of ants:Paratrechina steinheili, Solenopsis A, duringboth intervals. The expectedvalues are thefollow- Pheidolemoerens and SolenopsisB were foundmostly in ing: agriculturalland P(diurnal)= 0.75, P(nocturnal)= the first2.54 cm layerof litter;Paratrechina microps and 0.2 5; grasslandP(diurnal) = P(nocturnal)= 0. 5; forest Brachymyrmexminutus in the second layer; Anochetus P(diurnal)= 0.66, P(nocturnal)= 0.33. mayriand SolenopsisC in the third. Ants are predominantlydiurnal in agriculturalland and grassland.Paratrechina steinheili, Pheidole subarmata DAILY ACTIVITY.-Using the data on speciespresent on and Wasmanniaauropunctata can be consideredpredom- baitsput duringthe day and night,I performeda repli- inantlynocturnal if thenumber of activeworkers is taken

TABLE 4. Food itemscarried by ants in forest.

5 min obser- In- Ant/ In- In- Vegetal Food vation In- sect ant sect Feces/ Anne- sect Animal matter/ in Species periods Seeds sects parts parts larvae fungi lids eggs matter flowers crop Others I. melleus 119 24 38 8 12 1 2 7 73 23 M. ramulorum 80 1 8 39 30 5 22 7 28 5 296 41 P. steinheili 20 3 3 1 2 1 1 3 P. moerens 29 31 8 2 1 5 3 1 2 2 9 SolenopsisB 30 _ 3 5 _ 9 8 24 6 Total 278 1 69 93 41 6 50 5 8 40 15 393 82

288 Torres

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions TABLE 5. Ant microhabitatutilization in theagricultural land. S= sun; Sh = shade; N = night

Vege- Vege- Vege- Ground Ground Rock Rock tation tation Litter Litter Ground Rock tation Litter (S) (Sh) (S) (Sh) (S) (Sh) (S) (Sh) (N) (N) (N) (N)

A. kempfi 7 B. heeri 7 2 6 1 6 3 B. minutus 1 1 3 1 1 C. emeryi 3 1 1 5 1 1 C. nuda 19 1 12 5 6 3 3 C. venustula 1 1 1 2 C rimosus 1 - - M. ebeninum 3 2 1 2 M. floricola 1 - 1 3 3 O. brunneus 1 1 P. longicornis 1 1 P. steinheili 3 1 2 1 5 1 1 P. exigua 1 P. fallax 27 16 20 23 1 1 18 17 9 3 7 P. subarmata 24 29 13 20 3 10 40 28 1 SolenopsisA 2 S. geminata 38 26 27 23 14 16 25 15 5 T. melanocephalum 4 2 3 6 6 7 11 11 1 2 2 4 T. bicarinatum 1 2 3 3 4 W. auropunctata 4 1 7 2 5 6 3 5 6

intoaccount instead of presenceor absenceat a bait. also a nocturnalforest species. Daily activitydoes nothelp Camponotusustus is a nocturnalagricultural land species much in reducingoverlap in the forest. and Monomoriumebeninum is diurnal in grassland.Ants Ant activityis highlyinfluenced by temperature(Ber- tendto be nocturnalin theforest. Camponotus ustus is kelhamer1980, Whitfordet al. 1981) and measure-

TABLE 6. Ant microhabitatutilization in thegrassland.

Ground Rock Grass Ground Rock Grass (day) (day) (day) (night) (night) (night) A. kempfi 1 B. heeri 9 4 5 2 1 1 B. minutus 1 C. emeryi 8 5 5 2 1 C. nuda 17 16 5 4 4 C. venustula 1 C. rimosus 1 1 1 1 2 H. opacior 1 1 1 M. ebeninum 1 1 1 M. floricola 4 3 5 2 1 O. brunneus 6 1 2 Oxyepoecussp. ? 2 P. stigma 1 P. longicornis 9 5 1 2 4 P. steinheili 12 7 4 11 10 13 P. fallax 25 23 11 9 7 9 P. subarmata 27 8 2 23 12 3 SolenopsisA-C 2 2 S. geminata 34 23 4 25 15 6 S. louisianae 1 1 T. melanocephalum 2 2 4 2 4 T. bicarinatum 5 2 4 3 5 3 W. auropunctata 5 6 9 11 9 7

Ant Coexistence and Niches 289

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions TABLE 7. Ant microhabitatutilization in theforest. I placed 20 baits in the sunnymicrohabitats and 48 in the remaining microhabitats.S = sun; Sh = shade; N = night.

Vege- Vege- Vege- Ground Ground Rock Rock tation tation Litter Litter Ground Rock tation Litter Species (S) (Sh) (S) (Sh) (S) (Sh) (S) (Sh) (N) (N) (N) (N) B. minutus 1 3 1 3 1 C. ustus 3 I. melleus 2 4 1 7 13 22 5 6 1 4 17 P. stigma 2 2 1 P. microps 1 1 1 1 P. steinheili - 11 2 4 12 3 15 P. moerens 2 11 9 2 6 15 17 15 4 23 SolenopsisB 5 4 2 11 8 11 SolenopsisA 1 3 SolenopsisC I S. rogeri 1 T. littorale 1

mentsof temperaturetolerance can help explaindiurnal Grasslandants differ in toleranceto hightemperature activity.Tolerance to 45?C is an indicationof theamount (F = 19.83; df = 30, 221; P = 0.00). Temperaturein- of timea speciescan forageat high temperatures(Table fluencesallocation of resourcesbut thereis still overlap. 9). I performeda one-wayANOVA usinga logarithmic In the agriculturalland and grassland,tolerance to high transformationto make the variances homogeneous. temperaturesis similarin the fungus-growingants and Speciesdiffer significantly in tolerancein agriculturalland diurnal species stand 45?C betterthan nocturnalones (F = 24.01, df = 29, 228; P = 0.00). Species that tol- (Wilcoxontest, P = 0.01). eratehigh temperatures usually do not toleratelow tem- The ANOVA showsthat forest species differ in tem- peratureswell (Torres,unpublished). Several species are peraturetolerance (F= 24.98; df= 14, 115; P = 0.00). included in the temperature-nichedimension of other Notably,the forestspecies withstand high temperatures species. formuch shorter time relative to thespecies in nonforested

TABLE 8. Assignmentof species to thenocturnal (N), diurnal (D) or shared(S) categories.Diet type:1 = insectivorous,2 = liquid food,3 = granivorous,4 = fungusgrowing.

Species Agriculturalland Grassland Forest

Ga Ga Ga Brachymyrmexheeri 18.98b D 9.60b D2 Brachymyrmexminutus 3.48c S2 2.05c S Cardiocondylaemeryi 4.83b DI 1 1.90b DI Cardiocondylanuda 7.97b DI 21.30b DI Iridomyrmexmelleus 3.27c S2 Monomoriumebeninum 24. 16bD2 Monomoriumfloricola 4.89b D2 5.78b D2 Paratrechinalongicornis 4.60b D2 3.98b D2 Paratrechinasteinheili 0.01c S2 2.14' S2 20.54b NI Pheidolefallax 18.49bD1 14.20bDI Pheidolemoerens 25.93b NI Pheidolesubarmata 3.23c S1 0.01 S1 SolenopsisB 27.97b N2 Solenopsisgeminata 6.15b D3 2.11' S3 Tapinomamelanocephalum 12.32b D 0.29' S Tetramoriumbicarinatum 4.92b N2 0.00CS1 Wasmanniaauropunctata 0.20c S2 1.05cS2 Pooled data forall species 34.15b D 23.70b D 56.83b N

a df= 1 forall tests;b p < 0.05; c not significant. 290 Torres

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions Insectivorous ants ( grossland) 0 7-0 O brunneus n- 22 I nsectivorous ants (agricultural land) 7.0 _6-0/ . p punctato = 18I 6 a5, brunneus nz 18 L b* D

6S0 40 pt allax n- 140 d5-0T bicarinatum n= 47

025.0 psubarmata n= 43 0 20 424-5 - loqisianae n=27 00~~~~~~~~~~~~ exigua n=28 C,Iongiemeryi n 52 H4-0 P.ollox n=274 - CL10 nuda n 34 0)3.5 2/ _0 70 ._C ustus n-26 o o P subormato nl-149 o 25 UA. 2-0 -slouisionae n=27 40 0 .5 -c. emeryi n=58 ~~ ~~ ~~ ~~ ~~~~- ~~~-C. nuda n 31 00 I.G -.0 05 E Liquid food (grassland) . E 2 5 /. 50/. 7 5Y. 95*/.0 4-0 - 0 nP ~. p longicornis 23

0 P 7?0 - steinheii n=27 0 M. ebeninum n= 102 ~~~~~~~~~~~B. ~~~-I-Or heeri n= 7 6-5 ______M. floricola n,224 60 25/ 50 /. 75% 95/. 5.5 Liquid food (agricultural land) 5.0- ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 7 E~~~~~~~~~~~~~~

5.0 7 W auropunctota n.129 167 Other grassland species 0 n- ~ ~~ ~ ~ ~~~~~4p. longicornis 40 c.hrmosus n=45

- 2-0 1. me I'leus in~~~~~~~~~~~~~~~~~~.emn112~5 0V2-0- ~~~~~~~~~~M.smithi nl-67 uopunctao nofl68 4500 - l~~~~~~~~~~~W T.thbnXr natu Foetans2I8elusn |-

25*/ 50 ?/ 75/.4 95/. I 2F 5 . 50N y 7p5% 95%

1.0 I ~ ~ * ~ ~ * ~ ~~~~~I 6he - Fungus grower and granivorous od0n-ants EA(agricultural land) adi0 E 50 rimosus fl=60 0 Forest ants ~~~~~~~~~~~~~~~~~~~~~C. I. melleus n-97 M. smithi n. 109 0 4.4- N ~~~~~~~~~~~~~~~~~~~~~~S.geminata n =167 y350 p moerens nl-35 - ~ ~ ~~ ~ ~ ~ ~~~~~~~~~~~rP ste IInheIli flz8 3.0 2-0 U) ramularum M. nfl189 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 SolenopsisB n=31 i1.0 U~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I.50% 75/. ~~~~~~~~~~~~~~~~~~~~~~~~~~~25/. ~~~~ 95%

25/. 50 /. 75*/, 9 5/

FIGURE 1. Distributionof food sizes consumedby speciesin the threehabitats. Ninety-five percent, 75%o, 50%o, and 25%oof the itemsfall below the size indicatedon the ordinate.

AntCoexistence and Niches 291

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions TABLE 9. Mean amountof time tolerated by ants in theagricultural land (A. land); grassland(G. land); and forestat 45?C. Bars representspecies with homogeneous means (SNK procedure).See Table 1 forspecies code.

A. land SNK (min) G. land SNK (min) Forest SNK (min) Pa.m 7.9 An.m 9.1 Pacs 5.0 My.s 8.4 Hy.e 10.2 An.m 5.5 An.m 10.2 Cy.r 10.5 St.g 7.0 Cy.r 11.4 My.s 10.7 Am.s 8.8 Hy.e 11.8 St.1 14.1 Pa.m 9.7 St.l 11.1 Ox.s 14.5 So.C 10.6 Hy.o 12.1 So.C 15.5 St.r 10.9 So.C 12.3 Pacs 16.0 Hy.o 12.1 Pacs 15.7 Hy.o 16.8 So.B 13.4 Br.m 17.6 Phsu 17.0 Pa.s 14.6 Phsu 17.6 Br.m 18.5 Br.m 15.0 Ph.m 18.0 Od.b 18.7 Ph.m 16.2 Od.b 20.9 So.A 22.4 So.A 17.0 Ph.e 21.8 Ph.e 23.2 Ir.m 23.1 Ir.m 22.3 Pa.s 24.6 Camu 31.0 So.B 22.8 Te.b 24.8 So.A 24.6 Wa.a 26.7 Wa.a 25.7 Ro.s 26.8 Pa.s 26.2 So.B 31.8 Ta.m 33.3 Te.c 33.2 Te.b 34.2 Ta.m 42.2 Ph.f 40.8 Ph.f 42.9 So.g 46.1 Br.h 46.9 Ca.e 47.5 Co.g 47.6 Camu 49.2 Ca.n 48.1 Pa.l 50.8 Pa.1 54.7 Ca.n 54.2 Ca.e 56.6 Br.h 57.1 Cams 57.5 Mo.f 116.2 Mo.f 98.7 Mo.e 169.2 Ca.v 129.6 Mo.e 141.8

areas. However,the importanceof temperaturetolerance The null hypothesiswas rejected(G = 22.7, df = 1, P < in the forestshould not be overstressedbecause most 0.005), indicatingan advantage in arrivingfirst. This forestspecies are predominantlynocturnal and hightem- allowspoor competitorsto win some encounters,and will peraturesare rarelyexperienced in the forest,much less sway a competitivesituation on behalf of the weaker at night. competitor.This advantagetakes place because the first arrivalsrecruit and outnumberthe scouts from other INTERFERENCE COMPETITION.-I monitored409 interfer- species.Aggressive contacts with the speciesat the bait enceinteractions on tuna fishbaits, pooling grassland and seem to discouragescouts fromlaying odor trails,pre- agriculturalland observations.One species in each pair ventinglate-arriving species from recruiting workers. won in combat,regardless of environmentalconditions, I constructedan indexof wins-per-encounter in which exceptfor Pheidole fallax vs P. subarmataand P. fallax thenumber of encounterswon by a givenspecies over all vs Wasmanniaauropunctata. The sample for the latter the otherspecies was divided by the total numberof pair is too small to make any statisticaltests. Pheidole encountersin whichthe specieswas involved.This index fallax won diurnalcombats and P. subarmatanocturnal indicatesthe species'rank as a fighter(Table 10). Except ones (Fisher'sexact test,two-tailed, P = 0.02). Pheidole forMonomorium ebeninum and M. floricola,those species fallax won combatsin directsunlight under dry and wet that have theirniches included in that of anotherare conditions.Pheidole subarmata seems to be a betterfighter verypoor fighters.Many do not display any kind of in shade (Fisher'sexact test,two-tailed, P = 0.06). aggressivebehavior. I calculatedthe indexvalues fordif- There is a priorityeffect; ants arrivingto a bait first ferentenvironmental conditions and foundthat predom- have a greaterprobability of winninga combat.Data for inantlynocturnal species are betterfighters during the all the species were pooled and testedagainst the null night(x2 = 13.59, df = 1, P < 0.005) and in shade(x2 = hypothesisthat winningis independentof arrivaltime. 6.70, df = 1, P < 0.01). Becausepoor fightersare found

292 Torres

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions necessaryfor species coexistence. However, several species TABLE 10. Relativeindices of interference. use environmentalresources in similarways (e.g., Mono- Grassland/ moriumebeninum and M. floricola,Cyphomyrmex rimosus Species crops Forest and Mycocepurussmithi, emeryi and C. nuda, Cardiocondylaspecies and Pheidole Pheidolemoerens .50 fallax). Differencesin Solenopsisgeminata .22 dietand dailyactivity could facilitatecoexistence in grass- SolenopsisB .20 land ants.Species that use theenvironment similarly occur Monomoriumfloricola .20 here as well (e.g., Cardiocondylaemeryi and C. nuda, Monomoriumebeninum .10 Monomoriumebeninum and M. floricola,Paratrechina lon- Wasmanniaauropunctata .09 Pheidolefallax .08 gicornisand Brachymyrmexheeri, Cyphomyrmex rimosus Odontomachusbrunneus .08 and Mycocepurussmithi). Diet and litterutilization could Pheidolesubarmata .06 facilitateforest species coexistence. Paratrechinalongicornis .06 Overlap in differentniche dimensionswas indepen- Tetramoriumbicarinatum .06 dent,indicating a lack of Paratrechinasteinheili .02 .09 overlapcompensation. Repeated Tapinomamelanocephalum .02 patternsin the utilizationof the environmentare seen in Iridomyrmexmelleus .00 .21 these assemblages.Insectivorous ants in the agricultural Oxyepoecussp. ? .00 and grasslandcommunities forage mainly on theground. Brachymyrmexminutus .00 .00 Liquid feedingants in agriculturalland and grassland Brachymyrmexheeri .00 Hypoponeraopacior .00 foragemainly on vegetation.In the forest,most ant ac- Cardiocondylanuda .00 tivityis concentratedwithin the litter. There is greatover- Cardiocondylavenustula .00 lap in size of foodcarried to the nestby individualwork- Pheidoleexigua .00 ers.Any differencesamong speciesare furtherreduced as Paratrechinamicrops .00 food itemstoo large foran individualare overcomeby recruitingother individuals. In addition,forest ants tend to be nocturnaland ants in nonforestedareas are mostly in shaded microhabitats,good fightersin sunlightwin diurnal. moreconsistently in shade (paired t-test,arcsin transfor- AlthoughI founddifferences in temperaturetolerance mation, t = 2.43, df= 9, P < 0.05). whichare correlatedwith activitypatterns, there is also Aggressiveinteractions are rare in the forest.I wit- much overlapand manypatterns of nicheinclusion. The nessed27 combatson 464 baits (5.8%) in the forestas mainlynocturnal habits of the forestspecies minimize the comparedto 409 combatson 1056 baits (38.7%) in the importanceof toleranceto high temperaturesfor these nonforestedareas. There is no advantageto arrivingfirst species. at a bait in the forest(G= 0.33, df= 1, P > 0.50). Food limitationseems to be an importantfactor in the structuringof these ant assemblages.Interspecific DISCUSSION aggressionamong ants at concentratedfood sourcessup- portsthis contention. Two factorsthat affect the outcome Analysesof resourcepartitioning often have been based of these encounterswere discovered:microclimate and on thepredictions of simpleLotka-Volterra equations (but priority.The influenceof microclimateis best observed see Paine 1966; Caswell 1976; Simberloff1979; Sousa in interactionsbetween Pheidole fallax and P. subharmata, 1979). The main conclusionobtained from these simple but it is also evidentwhen one species' fightingabilities linearizeddifferential equations is that stable coexistence are comparedwith those of otherspecies. The priority of n species is attainedif the species are limitedby n effectobserved in combatsat tuna fishbaits is significant factors(Levin 1970). This equilibriumanalysis is exem- in theallotment of permanent food sources such as nectar- plifiedin thetheory of nichedifferentiation which predicts producingplants and honeydew-excretinginsects; nor- thatspecies must differ in orderto coexist.This approach mally weak competitorscan keep theirfood supply in has been successfulin explainingcoexistence in simple this manner.For evenlymatched species, like Monomo- communities.Nonetheless, it has failedto explaincoexis- tlumebeninum and M. floricola,priority effects are likely tencein fishes(Sale 1982), tropicalplants (Connell 1978) to be more influential.However, the priorityeffect does and vegetativedones (Harper 1977). not facilitatecoexistence in manyof the speciesthat have Nor does dassical nichedifferentiation theory entirely theirniche induded in those of otherspecies, because explainspecies coexistence in theseant communities.Ag- mostof the formerdo not exhibitaggressive behavior. riculturalland ants differin diet,daily activity, food size Levinset al. (1973), Wilson (1971), and Carrolland and macrohabitatutilization. These differencescould fa- Janzen(1973), foundpriority effects operating at several cilitatespecies coexistence in thiscommunity. By facilitate levelsin ant communities.They observed that established I mean that these factorscan help but might not be colonieshave a strongadvantage over incipientcolonies

Ant Coexistence and Niches 293

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions and foundingqueens. Queens of competitivelysuperior linearlyrelated to resourceabundance, cooperative hunt- speciescannot colonize an areaalready occupied by weak ing, multiplenutritional requirements, etc., predictco- competitorsbecause of predationor aggressionby the existenceof n speciesin k < n resources(Levins 1979, weakspecies. I haveobserved these priority effects in the Armstrongand McGehee 1980). The mentionedfactors ant communitiesstudied suggesting that they play an thatallow coexistenceof n speciesin k < n resourceshave importantrole in theorganization of thesecommunities. been foundin severalant communities(Wheeler 1936; In addition,ant aggressionis age dependent(Wheeler Sheppe 1970; Carrolland Janzen 1973; Whitfordand 1910, Carrolland Janzen 1973) pointingto anotheref- Ettershank1975; Davidson 1977b). fectof age structurein a competitiveenvironment. Antaggression was low in theforest compared to the ACKNOWLEDGMENTS nonforestedareas. This is in accordancewith the findings of Crowell(1968), Haskinsand Haskins(1965) and Thismanuscript was adaptedfrom my doctoral dissertation at Culver(1974), whofound that aggression is more prev- theUniversity of California(Berkeley). I would like to thank RichardLevins, Daniel Simberloff, Robert K. Colwell,Ana M. alentin structurallysimple habitats. Contrary to thesug- Marin,Roy Snelling,Donald Strong, Shahid Naeem, Herbert gestionthat species having their niches induded in those G. Baker,David Wake,Fabian Jaksic, Lucia Gonzgez, Lloyd of otherspecies should behave aggressively in orderto Goldwasser,Julio Cardona, Armindo Nuiiez, Ruben Rodri- coexist(Colwell and Fuentes1975), I foundthat these guez,Carlos Laboy, Harold L. Froler,James Ackerman and an anonymousreviewer for their help duringthe courseof this speciesusually do notbehave aggressively orare losers in work. interspecificcombats. Duringthis study I waspartially supported by a FordFoun- The currentfinding concerning the greater species co- dationFellowship. The Centerfor Latin American Studies, the existencerelative to apparentresource numbers is no sur- Universityof California,and theDepartment of Zoologypro- prise.Theoretical models dealing with feeding rates not videdtravel and computerfunds.

LITERATURE CITED ARMSTRONG R. A., AND R. McGEHEE. 1980. Competitiveexclusion. American Naturalist 115: 151-170. BERKELHAMER, R. C. 1980. Reproductivestrategies in ants: a comparisonof single-queenedversus multiple-queened species in the subfamilyDolichoderinae (: Formicidae). Ph.D. Thesis,University of California,Berkeley. BRIAN, M. V. 1956. Segregationof speciesof the ant genus Myrmica.Journal of Animal Ecology25: 319-337. , J. HIBBLE, AND A. F. KELLY. 1966. The dispersionof ant species in a southernEnglish heath. Journalof Animal Ecology35: 281-290. BROWN, W. L. 1973. A comparisonof the Hylean and Congo-West Africarain forestant faunas. In B. J. Meggers,E. S. Ayensu,and W. D. Duckworth(Eds.). Tropical forestecosystems in Africaand South America:a comparativereview, pp. 161-186. SmithsonianInstitution Press, Washington. CARROLL, C. R., AND D. H. JANZEN. 1973. Ecologyof foragingby ants. Annual Review of Ecologyand Systematics4: 231- 257. CASWELL, H. 1976. Communitystructure: a neutralmodel analysis.Ecological Monographs 46: 327-357. COLWELL, R. K., AND E. FUENTES. 1975. Experimentalstudies of the niche.Annual Review of Ecologyand Systematics6: 281- 310. CONNELL, J. H. 1978. Diversityin tropicalrain forestand coralreefs. Science 199: 1302-1310. CROWELL, K. L. 1968. Rates of competitiveexclusion by the Argentineant in . Ecology49: 551-555. CULVER, D. C. 1974. Speciespacking in Caribbeanand northtemperate ant communities.Ecology 55: 974-988. DAVIDSON, D. W. 1977a. Species diversityand communityorganization in desertseed-eating ants. Ecology58: 711-724. 1977b. Foragingecology and communityorganization in desertseed-eating ants. Ecology58: 725-737. EWEL, J., AND J. L. WHITMORE. 1973. The ecologicallife zones of Puerto Rico and the U.S. Virgin Islands. Forest Service researchpaper ITF-18, Rio Piedras,Puerto Rico. FUENTES, E. R. 1976. Ecologicalconvergence of lizard communitiesin Chile and California.Ecology 57: 3-17. HARPER, J. L. 1977. Populationbiology of plants.Academic Press,New York. HASKINS, C. P., AND E. F. HASKINS. 1965. Pheidolemegacephala vs Iridomyrmexhumilis in . Ecology46: 736-740. LEVIN, S. A. 1970. Communityequilibria and stability,and an extensionof the competitiveexclusion principle. American Naturalist104: 413-423. LEVINS, R. 1979. Coexistencein a variableenvironment. American Naturalist 114: 765-783. , M. L. PRESSICK, AND H. HEATWOLE. 1973. Coexistencepatterns in insularants. AmericanScientist 61: 463-472. LIEBERBURG, I., P. M. KRANZ, AND A. SEIP. 1975. Bermudianants revisited:the statusand interactionof Pheidolemegacephala and Iridomyrmexhumilis. Ecology 56: 473-478. MILLER, R. S. 1968. Conditionsof competitionbetween redwings and yellow-headedblackbirds. Journal of Animal Ecology37: 43-62. NATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION. 1979. Climatologicaldata: Puerto Rico and Virgin Islands. National ClimaticCenter, Asheville, N. C. PAINE, R. T. 1966. Food web complexityand speciesdiversity. American Naturalist 100: 65-75.

294 Torres

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions SALE, P. F. 1982. Stock-recruitrelationships and regionalcoexistence in a lotterycompetitive system: a simulationstudy. American Naturalist120: 139-159. SCHOENER, T. W. 1974. Resourcepartitioning in ecologicalcommunities. Science 185: 27-39. SHEPPE, W. 1970. Invertebratepredation on termitesof the Africansavanna. Insectes Sociaux 17: 205-218. SIMBERLOFF, D. 1979. Rarefactionas a distribution-freemethod of expressingand estimatingdiversity. In J. P. Grassle,G. P. Patil, W. K. Smith,and C. Taillie, (Eds.). Ecologicaldiversity in theoryand practice,pp. 159-176. InternationalCo- operativePublishing House, Maryland. SKAIFE, S. H. 1964. Las hormigas.Aguilar S. A., Madrid. SOUSA,W. P. 1979. Disturbancein marineintertidal boulder fields: the nonequilibriummaintenance of speciesdiversity. Ecology 60: 1225-1239. TAYLOR, B., AND S. F. ADEDOYIN. 1978. The abundance and inter-specificrelations of common ant species (Hymenoptera: Formicidae)on cocoa farmsin westernNigeria. BulletinEntomological Research 68: 105-121. TORRES, J. A. 1981. The organizationof ant assemblagesin PuertoRico: factorsinfluencing the diversityand coexistenceof species.Ph.D. Thesis,University of California,Berkeley. . 1984. Diversityand distributionof ant communitiesin PuertoRico. Biotropica16: 296-303. WHEELER, W. M. 1910. Ants. Columbia UniversityPress, New York. . 1936. Ecologicalrelations of ponerineand otherants to termites.Proceedings of the AmericanAcademy of Artsand Sciences71: 159-243. WHITFORD, W. G., AND G. EYrERSHANK. 1975. Factorsaffecting foraging activity in Chihuahuandesert harvester ants. Envi- ronmentalEntomology 4: 689-696. , D. J. DEPREE, P. HAMILTON, AND G. EYrERSHANK. 1981. Foragingecology of seed-harvestingants, Pheidolespp., in a Chihuahuandesert ecosystem. American Midland Naturalist105: 159-167. WILSON, E. 0. 1971. The insectsocieties. Harvard University Press, Cambridge, Massachusetts.

Ant Coexistence and Niches 295

This content downloaded from 166.4.81.212 on Thu, 25 Feb 2016 15:37:37 UTC All use subject to JSTOR Terms and Conditions