J. New YorkEntomol. Soc. 101(1);135-140, 1993 PHYSICAL AND BIOTIC CORRELATES OF POPULATION FLUCTUATIONS OF DOMINANT SOIL AND LITTER ANT SPECIES (HYMENOPTERA: FORMICIDAE) IN BRAZILIAN COCOA PLANTATIONS Jacques H. C. Delabie and Harold G. Fowler Secq:ao de Entomologia, CEPEC, CEPLAC, 45600 Itabuna, Bahia, Brazil, and Departamento de Ecologia, Institute de Biociencias, UNESP, 13500 Rio Claro, Sao Paulo, Brazil Abstract.—T\\crelation between weekly patternsofpercentoccurrenceofdominantantspe- cies in litterand soil samplesfroma Bahian (Brazil)cocoa plantation andclimaticfactorswas examined. No significant lagged correlationswere found forthe litterdominants, howeverthe soil dominant Acropyga cf paramaribensis showed significant lagged correlations with tem- perature, rainfall and leafflush, probably in response to increases in the populations oftheir mutualistic root mealybugs. Population fluctuationsofA. cf.paramaribensiswere found to be highlyassociatedwith thoseoftheothersoil dominant Tranopeltasp., which webelievetobe a predatorofthedominant and its mutualistic mealybugs. Because ofthe modular nature ofant colonies (Andersen, 1991), worker popula- tions vary more than do colony densities. These population fluctuations should be in response to resource availability, which should be strongly influenced by abiotic factors. Alternatively, population fluctuations could be due to interactions among members ofthe community, but probably still in response to resource availability. Ant communities are thought to be organized through competitive relationships (Fowler and Claver, 1991), and the occurrence of numerically dominant species necessarily limits resource availability for the remaining species ofthe community, as well as controls the upper limits ofspecies numbers (Tilman, 1982). Tropical ant communities have high faunal richness(Andersen, 1991). Ofthe 124 ground ant species recorded in a Bahian cocoa plantation, Solenopsis sp. A, Was- manniaauropunctata(Roger), Solenopsis{Diplorhoptrum)sp. B,and Carebarellasp. weredominantinthelitterstratum, andAcropygacf.paramaribensisBorgmeierand Tranopeltasp. were dominantsin the soil stratum (Delabie and Fowler, 1992). This strong dominance structure is not characteristic oftropical areas (Andersen, 1991), and the causes ofthese patterns are important components for an understanding of the organization ofcommunity structure (Tilman, 1982). The remainingant species ofthese cryptic communities were collected only occasionally, and, thus, their im- portance in structuring these communities may be limited. Here, we address the followingquestions: [1]doesweatherhaveanyrelationtothepopulationfluctuations ofthedominantantspecies?;and[2]dostronginteractionsexistamongthedominant ant species? Either or both ofthese could produce changes in community structure. 136 JOURNAL OFTHE NEW YORK ENTOMOLOGICALSOCIETY Vol. 101(1) METHODS AND MATERIALS FromJune, 1986 to December, 1987, inanarea of0.83 haon the Cocoa Research Station in Itabuna, Bahia, Brazil, 5115x15x15 cm weekly soil samples were collected immediately below the litter layer. Each sample was taken 1.5 m from the trunk ofone of 500 randomly chosen numbered trees. Ants were immediately ex- mm tracted by hand sorting, and afterwards two successive sifters (2 and 1 mesh size) were used to extract ants not visually detected. Ants were preserv'ed in 70% alcohol for subsequent identification in the laboratory. Littersampleswerecollectedatirregularintervalsfrom February, 1987toOctober, 1988, in the same experimental area, using the same sampling procedure. For each sample, a randomized 1 m- of litter near the trunk of a numbered tree was also collected, and its depth measured. These samples were extracted in a Berlese funnel for48 hours, andthen preserved in 70%alcohol and identifiedasinthesoilsamples. For both soil and litter samples, we recorded species presence. Data were then transformed into percentages (number ofsamples per species/51 weekly samples). For both sampled strata, we compared only the dominant species, defined through speciesrank-abundanceplots, usingacriterionof10%occurrenceasaminimalvalue (Delabie and Fowler, 1992). Wecorrelatedtheweeklypercentageofsamplescontainingdominantspeciesagainst data from the nearby Cocoa Research Center weather station. Data were smoothed bya 3 point movingaverage, and were then successively lagged forcorrelations. We regressed monthly percentage occurrence ofthe dominant soil species against each other. RESULTS Dominant litter species were Solenopsis sp. .A (44.03%), S. (Diplorhoptrum) sp. B (13.82%), B'. auropunctata (10.07%) and Carebarella sp. (10.07%), while for soil, dominant species wereA. cfparamaribemis (58.07%) and Tranopelta sp. (13.17%). No significant correlations were found for any ofthe dominant species with tem- perature or precipitation (Fig. 1). However, a strong peak occurrence frequency of Solenopsis sp. A was evident at the end ofthe summer (April) (Fig. 1). The other dominant ant species showed no apparent seasonal trends (Fig. 1). Similarcomparisonswereperformedwiththedominantsoilspecieswith leafflush andclimaticconditions(Fig. 2). Afterdata smoothing,A. cfparamaribensispercent occurrence was positively correlated with precipitation at a 5 week lag {r = 0.60; P = 0.00 while temperature and leafflush were positively correlated with ant pop- 1), ulations at a 12 week lag(r= 0.52; P= 0.001 and r= 0.51; P= 0.001, respectively). By regressing the numbers ofeach ofthe two species per month for two species (Fig. 3), a strong relationship was found: y = 0.33x - 2.88 (r^ = 0.64) where y = percent of.J. cf paramaribensis in the monthly samples, and x = percent of Tranopelta observed in the samples during the same period. DISCUSSION Two of the dominant litter species are facultatively dependent upon honeydew (W. auropunctata and Solenopsis sp. A). Carebarella sp. and 5. {Diplorhoptrum) sp. 1993 SOILAND LITTER ANTS 137 °C mm Fig. 1. Percent occurrence ofthe dominant ant species in Berlese litter samples over the study period, and correspondingairtemperatureand rainfall data. 138 JOURNAL OFTHE NEW YORK ENTOMOLOGICAL SOCIETY Vol. 101(1) % Aero pygo spp TronopeIfo sp % mm Fig. 2. Percentoccurrenceofthedominantantspeciesinsoilsamplesoverthestudyp>eriod, and correspondingairtemp>erature and rainfall data. 1993 SOIL AND LITTERANTS 139 Percent of Acropyga Fig. 3. The relation between the percent occurrence ofthe two dominant soil ant species overthe samplingperiod. B are both strictlycarnivorous(Fowleretal., 1991), and foragingpopulations would dependon thetypeandabundanceofpreypresent, which wasapparentlynotdirectly influenced by climatic conditions. We suggest that climate may indirectly influence the populations ofA. cf para- maribensis and its associated mealybugs. Climate influences both plant growth and leaf flushing, and we suspect that the mealybugs may have a greater physiological dependence upon the root’sgrowth andturgorpressure, thuslimitingtheavailability offoodresources(Delabieetal., 1991).Thismayexplainthe 12weeklaginpopulation response to temperature and leaf flush. The 5 week lag detected for population response to rainfall may indeed be an indication that humidity induces a quick population response ofthe associated mealybugs, and a corresponding increase of ant populations (Zanetti, 1992). We have no information on the interrelations between A. cf paramaribensis and Tranopelta sp. However, we suggest three possibilities to explain the positive cor- relations between population frequencies. The first possibility is that both depend upon mealybugs as a food resource. However, ifmealybugs, or mealybug root sub- strate were limiting factors, this correlation should be negative to evidence compe- tition. The second possible explanation is that Tranopelta preys uponAcropyga, but we have no evidence that this occurs. The data are, nevertheless, very reminiscent oftypical predatorbehavioral functionalresponses. Thethirdhypothesisisthatboth species are facultatively dependent, or that at least one species depends upon the other for either food or protection. This alternative was chosen in our regression model. Tranopelta sp. is a possible parent ofthe Solenopsidini tribe (Myrmicinae), 140 JOURNAL OFTHE NEW YORK ENTOMOLOGICAL SOCIETY Vol. 101(1) which is primarily predaceous, although some species will facultatively tend ho- mopterans (Fowler et al., 1991). However, species of Acropyga (Formicinae) are entirely dependent upon root mealybugs as a food resource, formingone ofthe true mutualistic relationships among insects (Fowleret al., 1991). Therefore, we propose that Tranopelta may indeed protect colonies ofAcropyga, and extract a tribute in eitherAcropygaworkersorbrood,orinsymbioticrootmealybugs.Indeed, Tranopelta sp. is encountered much less than expected at both carbohydrate and protein sub- terranean baits (Fowler and Delabie, 1992), providing further indirect evidence for their facultative dependence. Obviously, controlled laboratory studies are needed to test these competing hypotheses. ACKNOWLEDGMENTS Studies were supported through the Conselho Nacional de Desenvolvimento Cientifico e Tecnologico (CNPq) of Brazil, Grants 500185/88-3 and 300171/88-9, and CEPLAC. Edna Ferronato provided support in field work, and Irene Mauricio Cazorla assisted in the data analysis. We thank Marcos Antonio Pesquero, Marcelo Schlindwein and Sofia Campiolo for their comments, and the Fundagao para o Desenvolvimento da UNESP (FUNDUNESP) for publication support. LITERATURE CITED Andersen,A.N. 1991. Parallelsbetweenantsandplants:implicationsforcommunityecology. Pages 537-558 in:C. R. Huxleyand D. E. Culver(eds.),Ant-PlantInteractions.Oxford University Press, Oxford. Delabie, J. H. C. and H. G. Fowler. 1992. Soil and littercryptic ant assemblages ofBahian cocoa plantations. Pedobiologia (in press). Delabie, J. H. C., J. E. Mantovani and I. Mauricio. 1991. Observances sobre a biologia de duasespeciesdeAcropyga(Eormicidae,Eormicinae,Plagiolepidini)associadasarizosfera do cacaueiro. Rev. Brasil. Biol. 51:184-192. Eowler, H. G. and S. Claver. 1991. Leaf-cutterantassemblies: effectsoflatitude, vegetation, and behaviour. Pages 51-59 in:C. R. Huxleyand D. F. Culver(eds.), Ant-Plant Inter- actions. Oxford University Press, Oxford. Fowler,H.G.andJ.H.C. Delabie. 1992. Resourcepartitioningamongepigaeicandhypogaeic ants (Hymenoptera: Formicidae) ofa Brazilian cocoa plantation. Ins. Soc. (in press). Fowler,H. G., L. C. Forti,C. R. F. Brandao,J. H.C. Delabieand H. L.deVasconcelos. 1991. Ecologia nutricional de formigas. Pages 131-223 in: A. R. Panizzi and J. R. P. Parra (eds.), Ecologia Nutricional de Insetos. Editora Manole, Sao Paulo. Tilman, D. 1982. Resource Competition and Community Structure. Princeton University Press, Princeton. Zanetti,E. S. 1992. Caracteristicasdascomunidadesdeformigas(Hymenoptera: Eormicidae) esuasintera9oescomhomopterosempomaresdecitricos.M.S.thesis,UNESP,Botucatu, SP, Brazil. Received 8 January 1992; accepted 20 May 1992.