Journal of Natural History, 2002, 36, 2211–2227 Foraging ecology of the giant Amazonian ant Dinoponera gigantea (Hymenoptera, Formicidae, Ponerinae): activity schedule, diet and spatial foraging patterns VINCENT FOURCASSIE´† and PAULO S. OLIVEIRA‡* †Laboratoire d’Ethologie et Psychologie Animale, UMR CNRS N°5550, Universite´ Paul Sabatier, 118, route de Narbonne, F-31062 Toulouse Cedex 4, France ‡Departamento de Zoologia, Universidade Estadual de Campinas, C.P. 6109, 13083-970 Campinas SP, Brazil (Accepted 20 May2001) Thisstudyprovidesadetailedaccountofthenaturalhistoryandforagingbiology of the ponerine ant Dinoponera gigantea in a rainforest in north Brazil. The speciesnestsonthegroundandthecoloniescontain70–96workers.Antactivity isnegativelycorrelatedwithtemperature,andismoreintenseatdawnanddusk. Foragers leave the nest independently and search for food individually on the leaflitter,withinca10maroundthenest.Workersareopportunisticfeedersthat collectseedsandfruits,andhuntforlivepreyaswellasscavengefordeadanimal matter.Thedryweightoffooditemsrangesfrom<10mg(spiders,insectparts) to >400mg (seeds, fruits). There is no nestmate recruitment during the search for orretrieval offood,irrespective offood type and size. Foragers have a high directional(cid:142)delity,andantsfromneighbouringcoloniesmayengageinritualized territorialcontestsattheborderoftheirforagingareas.Theforagingecologyof D.gigantea iscompared with other ponerine specieslivingin tropicalforests,as well aswith otherant groups showingsimilar behavioural patterns. Keywords:Activity rhythms, ants, Dinoponera, Formicidae, Ponerinae, spatial foraging patterns, territorialcontests. Introduction Ants outnumber all other terrestrial organisms and occur in virtually all types of habitats (Wheeler, 1910). The dominance of ants is particularly conspicuous in the tropical region, especially in rainforests, where they account, together with termites, for nearly one-third of the entire animal biomass (Fittkau and Klinge, 1973). Foraging ants may search for solid or liquid food on the ground and/or on foliage, and the dietary requirements across diVerent species within the family *Author towhom correspondence is addressed; e-mail: [email protected] JournalofNaturalHistory ISSN0022-2933print/ISSN1464-5262online©2002Taylor&FrancisLtd http://www.tandf.co.uk/journals DOI:10.1080/00222930110097149 2212 V. Fourcassie´ and P. S. Oliveira Formicidae is extremely diverse (Carroll and Janzen, 1973; Ho¨lldobler and Wilson, 1990). Ants may also vary widely in the strategies they use to gather food, and foraging modes may range from solitary hunting in which there is no co-operation duringsearchandfoodretrieval,to varyinglevelsofco-operativeforagingmediated by diVerent degrees of recruitment communication between nestmates (Ho¨lldobler and Wilson, 1990). Because the foragers usually depart from a (cid:142)xed nest location, antsareregardedas usefulorganismstotestanumberofhypothesesaboutforaging strategies. According to Traniello (1989), a full understanding of an ant colony’s foragingsystemrequiresthedeterminationofboththeindividualandsocialcompon- ents of the foraging behaviour, as well as of the ecological setting in which the colony occurs (i.e. its resource and competitive environment). Therefore more nat- uralhistoryinformationandquantitativedataareneededabouttheforagingcharac- teristics of diVerent species before we can properly assess the selective pressures underlying observed ant foraging systems. AvarietyofforagingstrategiesoccursacrossdiVerentantspeciesintheprimitive subfamilyPonerinae(PeetersandCrewe,1987).Foragingmodesrangefromsolitary to group hunters, and from specialist to generalist predators (e.g. Ho¨lldobler, 1982; Fresneau, 1985; Maschwitz et al., 1989; Lachaud, 1990; Duncan and Crewe, 1994; Leal and Oliveira, 1995). Although generally regarded as predatory ants, ponerine species may also feed on extra(cid:143)oral nectar, homopteran honeydew, secretions from lepidopteranlarvae, as well as lipid-rich seeds(e.g. Horvitz,1981; Ho¨lldobler,1985; DeVries, 1991; Oliveira and Branda˜o, 1991; Pizo and Oliveira, 1998; Del-Claro and Oliveira, 1999). Species in the Neotropical ponerine genus Dinoponera are among the largest known ants. Workers attain a size of 3–4cm in length (Kempf, 1971; Paiva and Branda˜o, 1995). They nest on the ground and are distributed throughout South Americainseveralhabitattypes,rangingfromaridsavannasto rainforests(Kempf, 1971). Dinoponera species are queenless and reproduction is performed by mated workers (Dantas de Araujo et al., 1990). Colony foundation may occur by (cid:142)ssion (Overal, 1980). Despite their impressive size and local abundance, little is known about the foraging ecology of Dinoponera ants. Although the few available reports indicatethattheantsaregeneralistpredatorsonlitterarthropodsandsnails(Fowler, 1985; Paiva and Branda˜o, 1995), to date there is no quantitative account on the diet or foraging behaviour of Dinoponera. This paper presents a detailed (cid:142)eld account on the foraging ecology of the Amazonian ant Dinoponera gigantea. Previous observations on this species indicate that workers forage individually and show some (cid:142)delity to a foraging area (Fourcassie´ et al., 1999). We here provide further qualitative and quantitative data onthenaturalhistoryandforagingbiologyofD.gigantea,withemphasison colony demography, activity rhythms, dietary requirements and spatial foraging patterns. Materials and methods Field work was carried out during December 1999 (end of the dry season) in a secondary rainforest site located at the Fazenda Vito´ria, near Paragominas (2°59êS, 47°31êW), Stateof Para´,northBrazil. The physiognomyof thevegetationconsisted of trees and palms (up to 30m tall) and scattered understory shrubs (#1–2m tall) growingovera thicklayerof leaflitter. Seventeennestsof D.gigantea were marked inthestudyarea,andtheirmainexternalcharacteristicswererecorded(i.e.location, numberanddiameterof nestentrances).Attheendofthebehaviouralobservations Foraging ecology of Dinoponera ants 2213 four nests were excavated to determine their depths and the demography of the colonies. Ant voucher specimens are deposited in the Museu de Zoologia da Universidade de Sa˜o Paulo (MZUSP). Activity schedule For three colonies (Nos 9, 10, 12) all individuals seen outside the nests during the course of the study were individually marked on the thorax and/or gaster with a distinct colour code, using dots of enamel paint (Testors Co., Rockford, USA). The nests of these colonies were 30–44m apart from each other. Foraging rhythms at colonies Nos 9 and 10 were monitored continuously from dawn (6.00a.m.) to dusk(6.00p.m.).Duringtheobservationperiodallworkers(markedandunmarked) exiting or entering each nest were recorded, and air temperature was monitored at 1-h intervals. The duration of foraging trips by marked workers was calculated based on their departure and arrival times. Short trips by workers engaged in maintenance activities were also recorded (i.e. removal of nest refuse, removal of sticksandleavesfromnestentrance).Althoughreturningworkerstransportingfood itemswerenotdisturbedtoavoidalteringtheirforagingactivity(seeStadling,1978), some of the retrieved items were large enough to be identi(cid:142)ed on sight. Survey of food items The food items retrieved by D. gigantea were surveyed by removing them from the mandibles of returning foragers from any of the 17 marked nests, but mainly fromnestsNos9,10and12.Thisprocedureallowedthecollectionofalargenumber of food items. In cases where the removal of the item was avoided (see above), the identi(cid:142)cation of the food was included in the survey. Food items were preserved in 70% alcohol and brought to the laboratory for more detailed identi(cid:142)cation. The items were then kept in the oven at 60°C for 24h, and their dry weights were measured with a Mettler H51Ar analytical balance. Spatial foraging patterns In order to assess the extent of the foraging range of the colonies and to investigate the spatial pattern of individual foragers, marked ants were followed as soonas theyexitedanestby placing consecutively numbered(cid:143)agsalongtheirpaths at 1-min intervals (Turchinet al., 1991).The position of the (cid:143)agswas then mapped bymeasuringtheirdistancesrelativetotworeferencepoints,andbyusingtriangula- tion formulae. Ants were followed for up to 15min. Preliminary observations have shown that the maximum distance from the nest was generally reached within this period. We measured the spatial specialization of individual foragers by computing themeanvectorof thedistributionof the azimuthrelativeto thenestofall the(cid:142)xes composing their paths (Batschelet, 1981). The statistical signi(cid:142)cance of the mean vector was assessed by using the Rayleigh test. A signi(cid:142)cant vector indicates that the(cid:142)xesare not randomlydistributedaroundthenest, butare insteadconcentrated in a narrow angular sector. Results Natural history and demography Except for one isolated nest, all other nests of D. gigantea were located at the base of trees (N=12), palms (N=2) or thick lianas (N=2). The nests’ external 2214 V. Fourcassie´ and P. S. Oliveira appearance was conspicuous on the forest (cid:143)oor because their immediate vicinity was usually surrounded by yellowish soil particles due to excavation by the ants, which however, did not form a mound. Each nest had one to eight entrances (mean±SD=3.9±2.3; N=17) of 3–8cm in diameter, which were 3–250cm apart from each other. The four excavated nests were shallow (#40cm deep), with chambers ca 3-cm high and 20-cm wide. When in close proximity to each other (<10cm apart), theentrancesnormally mergedintoa singlegallery ca 15cm below the soil surface, which led to the nest chambers. On the other hand, nest entrances which were 40–250cm apart had no connection underground, indicating that D. gigantea hasalso polydomousnestinghabits(i.e.colonyoccupies morethanone nest; seeHo¨lldoblerandWilson,1990). The demographic dataof the colonies from the excavated nests are shown in table1. The colonies were reproductively active, and contained a signi(cid:142)cant amount of brood during the study period. The marking procedure revealed that 27–40% of the workers were engaged in activities outside the nest (i.e. maintenance, or foraging). Activity rhythms and diet The activity rhythms of monitored colonies was negatively correlated with tem- perature (colony No.9: r =-0.83, P<0.01; colony No.10: r =-0.78, P<0.01). s s Ant activity presented a bimodal distribution and was more intense at dawn and dusk,justbeforeandaftertheperiodofmaximumtemperaturefortheday((cid:142)gure1). However, since ants were also seen returning to the nest at sunrise, and leaving at sunset, some activity presumably also occurred during the night period. Foraging activity during the 12-h sampling was performedby 23–24% of the workers ineach colony, as revealed by the records of individual marked ants in proportion to the total number of ants per colony (determined after nest excavations). Ants typically leave the nest independently and search for food individually within a radius of ca 10m around the nest (see below). Qualitative observations, however, suggest that theants’foragingrangemayoccasionallysurpassthislimit,sinceantswereattracted to sardine baits placed ca20m from their nests. Foragers usually walk towards one direction for 15–20min, after which they begin to search around for food within a limited area. Such foraging trips could last up to 3h, but ants usually found prey items within 30–60min ((cid:142)gure2). In both colonies only 10% of the foraging trips weresuccessfulduringthe12-hsampling.Brieftrips(45min)aroundtheimmediate vicinityofthenestwereveryfrequentandconsistedmostlyofmaintenanceactivities Table1. Composition of four Dinoponera gigantea colonies, determined by excavation at theendofthebehaviouralobservationscarriedoutinsecondaryAmazonianrainfor- est,north Brazil,in December1999. Colony No. of No. of No. of No. of No. of No.ofworkers code workers pupae larvae eggs males marked (%) 9 70 15 18 0 1 25 (35.7) 10 95 38 5 0 8 38 (40.0) 11 75 27 19 6 2 – 12 96 5 7 11 0 26 (27.1) Coloniescontained84.0±13.4workers(mean±SD).ExceptforcolonyNo.11,whereno workers were marked,allworkers engaged inactivitiesoutsidethe nest(i.e.maintenance, or foraging) were marked. Foraging ecology of Dinoponera ants 2215 Fig.1. Activity rhythm of two Dinoponera gigantea colonies in a secondary rainforest in north Brazil. Ant activity is more intense at dawn and dusk hours, before and after maximum temperature for the day. such as the removal of nest refuse, or of sticks and leaves from nest entrances ((cid:142)gure2). Dinoponeragiganteaforagesexclusivelyontheground,neversearchingonplants. There was no evidence of recruitment communication between nestmates in the search for or during retrieval of food, irrespective of the type of prey. Foraging workers are opportunistic feeders that collect plant resources, and both hunt for live prey as well as scavenge for dead animal matter (table2). The ants included a wide array of food types in their diet, from seeds and fruits (22% of all retrieved items) to a taxonomically diverse assemblage of litter-dwelling organisms (table2). The size spectrum of prey items was also extremely wide ((cid:142)gure3), and their dry weightsrangedfrom<10mg(e.g.ants,spiders,insectparts)to>100mg(e.g.fruits, crickets, snails). Vismia fruits (Clusiaceae) and Inga seeds (Leguminosae) were the heaviest food items retrieved by the ants (>400mg). Individual foragers were able to carry aloft small- to medium-sized food items, and drag large ones through the leaf litter. Spatial foraging patterns A total of 67 tracks from 36 individual ants were recorded (colony No.9: 21 tracks of 13 ants; colony No.10: 24 tracks of 13 ants; colony No.12: 22 tracks of 10 ants). Figure4 shows the foraging range of each colony based on all tracks recorded. Except for colony No.9, there was no directional bias in the foraging 2216 V. Fourcassie´ and P. S. Oliveira Fig.2. Frequency distribution of trip duration relative to diVerent activities performed by workersof Dinoponeragigantea in a Brazilian rainforest site. Although foragingants may be away from the nest for up to 3h, successful foragers usually return after 30–60min of searching. Data are based on continuous 12-h observations at colony Nos 9 and 10,from 6.00a.m. to6.00 p.m. Two successful foragers from each colony are not included in the graphs because the duration of their foraging trips could not berecorded. Foraging ecology of Dinoponera ants 2217 Table2. List of food items collected by workers of Dinoponera gigantea in Amazonian rainforest in north Brazil, in December1999. No.ofrecords (%), No.oflive animal Taxonomicidentity offood item n=73 items prey Fungi Basidiomycetes(part ofmushroom) 1 (1.4) Angiospermae Seed orfruit 16 (21.9) Insecta Orthoptera 8 (10.9) 7 Hymenoptera Vespidae 1 (1.4) 1 Formicidae 3 (4.1) 3 Isoptera 1 (1.4) 1 Coleoptera Adult 2 (2.7) 2 Larvae 2 (2.7) 2 Odonata 1 (1.4) 1 Blattodea 4 (5.5) 4 Homoptera Cicadellidae 1 (1.4) Cicadidae 1 (1.4) 1 Hemiptera Pentatomidae 1 (1.4) 1 Lepidoptera Adult 1 (1.4) Pupae 2 (2.7) 2 Larvae 2 (2.7) 2 Arachnida Araneae 2 (2.7) 2 Opiliones 2 (2.7) Chilopoda Scolopendromorpha(partofbody) 1 (1.4) Diplopoda Polydesmida 1 (1.4) Turbellaria(part ofbody) 1 (1.4) Gastropoda Pulmonata 3 (4.1) Parts ofarthropods 13 (17.8) Unidenti(cid:142)ed 3 (4.1) Fooditemswereidenti(cid:142)edbyremovingthemfromreturningforagersofseveralcolonies, but mostly from colony Nos 9, 10 and 12. Collections were not made during systematic surveys ofantactivity,soas not toalter the ants’foragingrhythm. eVort of the colonies. Within the 15-min observation period ants travelled ca 6m (5.8±3.0; N=67),at amaximumdistanceof 12.3mfromtheirnests((cid:142)gure4).The paths of individual ants that were followed at least twice during the study period are represented in (cid:142)gure5 by distinct line patterns. The average period of time elapsed between two recorded tracks of the same ant was 3.5 days (range: 30min to9days).Antsrepeatedlyvisitedthesamezoneandshowedahighsectorial(cid:142)delity. This tendency is seen in the high signi(cid:142)cance of the mean vectors of the azimuths of the (cid:142)xes of their paths ((cid:142)gure5). 2218 V. Fourcassie´ and P. S. Oliveira Fig.3. Frequency distribution of the dry weights of food items (N=66) retrieved by Dinoponeragiganteaforagersinasecondary rainforest,north Brazil. Territorial behaviour When two ants from diVerent colonies met at the border of their foraging areas they engaged in a ritualized territorial contest that could last nearly 30min. During such agonistic encounters the ants usually faced each other frontally and locked their mandibles together ((cid:142)gure6A). While locked to one another the ants elevate the anterior part of their bodies, vigorously antennate each other’s head, and constantly kick one another with the forelegs. Eventually one of the ants became dominant and stood over its opponent, which was dragged away. As the contest escalated the dominant ant was observed to bite her subordinate on the head, and direct the tip of the gaster against the opponent’s body ((cid:142)gure6B). The ants may thenhave foughtbrie(cid:143)y, with thesubordinateantwalkingawayafterbreakingfree. No injury to either ant was detectable after such contests. Discussion The period in which ants can be active is largely determined by the species’ physiologicalproperties,inparticularbytheirtolerancelimitswithrespecttotemper- ature and humidity oscillations in the environment (Ho¨lldobler and Wilson, 1990). The activity rhythm of D. gigantea was negatively associated with temperature and followed a bimodal pattern in which most activity is con(cid:142)ned to early morning and late afternoon, with a marked decline around midday. Such an activity pattern is commonly seen in warm environments, and is also exhibited by other ponerine specieslivingintropicalforests(Le´vieux,1977;DejeanandLachaud,1994;Duncan and Crewe, 1994; Passera et al., 1994). In general, the colonies of D. gigantea distributed their foraging eVort fairly evenly around their nest. Some sectors, however, were visited less by the ants and this may be due to heterogeneities in the distribution of food resources (Levings, Foraging ecology of Dinoponera ants 2219 Fig.4. Maps showing the spatial foraging patterns of the three colonies of Dinoponera gigantea studied in a Brazilian rainforest. The hatched area around all the recorded tracksrepresentstheforagingrange.Thepositionofthenestentrancesisindicatedby a black square (note that colony No. 10 had two nest entrances). For each colony a circulargraph shows thedistributionofthe directionofthelast (cid:142)xofthe tracks.The radius of the circle corresponds to unity. The mean vector of the distribution is signi(cid:142)cant for colony No. 9 only (Rayleigh test). The frequency histograms at the bottom give the distance tothe nest ofthe last (cid:142)xofeach ofthe tracks. 1983),ortocompetitioneVectduetoneighbouringcolonies(seebelow).Dinoponera gigantea typically exhibits an individual foraging strategy (Beckers et al., 1989). Foragers independently hunt on live ground-dwelling organisms, and search for plant resources and dead animal matter within the leaf litter. This pattern has also been reported for several other ponerine species (Fresneau, 1985; Lachaud, 1990; Duncan and Crewe, 1994; Passera et al., 1994; Ehmer and Ho¨lldobler, 1995). However, as opposed to some solitary-hunting ponerines that may co-operate in preyretrieval (Ho¨lldobler, 1984;Oliveira andHo¨lldobler, 1989; Pratt,1989; Dejean et al., 1993), no recruitment behaviour has ever been observed in D. gigantea while searching for or retrieving food. Foragers depart from their nest repeatedly in the same direction and prospect for food within a restricted sector around their nest. Someindividualspersistedvisitingthesamelimitedareaforperiodsofupto 9days. This result complements that of Fourcassie´ et al. (1999) showing that homeward D.gigantea foragersusethesameroutethroughthevegetationovera3-weekperiod. Occasional observations of foragers returning with food items show that they spend less time in the nest than unsuccessful foragers, and that they return directly to and search at the location of their last food (cid:142)nd (see also Duncan and Crewe, 1994). Harkness and Maroudas (1985) and Deneubourg et al. (1987) have shown 2220 V. Fourcassie´ and P. S. Oliveira withtwodiVerentmathematicalmodelsthatthissimplespatialreinforcementprocess can lead at the individual level to a high spatial (cid:142)delity and at the colony level to a partitioningof the foragingground among workers. Sucha spatial foragingpattern hasbeendescribedinseveralantspeciessearchingsolitarilyfordispersedfooditems (Pogonomyrmexmaricopa:Ho¨lldobler,1976;Cataglyphisbicolor:Wehneretal.,1983; Schmid-Hempel, 1987; Pachyconyla apicalis: Fresneau, 1985; Ocymyrmex velox: Wehner, 1987; Formica schaufussi: Traniello, 1988; Paltothyreus tarsatus: De´jean et al., 1993a; Hagensia havilandi: Duncan and Crewe, 1994; Odontomachus bauri: EhmerandHo¨lldobler,1995;Messorarenarius:Warburg,1996).Solitarilysearching antscan relyexclusively on pathintegrationor on spatial memory to orient intheir environment. In heterogeneous environments with a dense canopy cover such as tropical forests, navigation by path integration is likely to be diYcult to implement sincethecelestialcuesnecessarytointegratethedirectionalcomponentsofthepaths areonlyvisibleintermittently.Thereforeantshavetonavigatebyusingtheirmemory ofthevisuallandmarksencounteredalongfamiliarroutes(Baader,1996;Fourcassie´ et al., 1999). The directional (cid:142)delity observedinthese antsmaythusbe constrained bythecostofbeinglostwhentheywanderawayfromtheirfamiliarsector.According to Wehner (1987), a forager ant is the less likely to abandon the direction of a preceding unsuccessful trip the more successful foraging trips it has achieved in the samedirection.Solitarysearchingantswouldconsequentlydevelopdirectional(cid:142)del- ity only if they have a high rate of reinforcement (i.e. if their foraging eYciency is high).ThelowforagingeYciencyobservedinD.gigantea coloniessuggests,however, that the cost of being lost may be strong enough a constraint to prevent the ants fromvisitingunfamiliarneighbouringsectors.Navigationalconstraintsmaythusbe ultimatelymoreimportantthanecologicalconstraintsinshapingthespatialforaging pattern of individual foragers (Fewell, 1990). External activities related to nest maintenance and foraginginvolved 27–40% of the D. gigantea workers, whereas in the African ponerine Hagensia havilandi such tasks are performed by 60–77% of the worker force (Duncan and Crewe, 1994). The proportion of workers involved in foraging was also smaller in D. gigantea (23–24%) than in H. havilandi (#40%; Duncan and Crewe, 1994). Foraging eYci- ency, as expressed by the proportion of foraging trips resulting in food retrieval, was lower in D. gigantea (10%) than in H. havilandi in Africa (37%; Duncan and Crewe, 1994), Ectatomma ruidum in Mexico (12–19%; Lachaud, 1990) and Odontomachus bauriinPanama (28%;Ehmer and Ho¨lldobler, 1995).Three factors, however,needtobetakenintoaccountinthesecomparisons.First,Lachaud(1990) has shown that foraging eYciency varies seasonally and that during the dry period E.ruidum’saveragesuccessratedecreasesfrom19to12%.Second,inthedryseason E. ruidum collects liquid food on plants, and if only solid food is considered the species’ foragingeYciency drops to 10% (Lachaud, 1990), as alsoreported here for Fig.5. MapsshowingthetracksofDinoponeragiganteaforagersthatwerefollowedatleast twice during the study period. Each line pattern corresponds to a diVerent ant. The labelsonthesideofthelinesindicatethedateatwhichthetrackswererecorded.The nestsarerepresentedbyablacksquare.Themain treesaresymbolizedbygreycircles whose diameter is roughly proportional to that of the trees. The circles on the right show the mean vector of the azimuths of the (cid:142)xes composing the paths of each ant. Theradiusofthecirclecorrespondstounity.Allvectorsarehighlysigni(cid:142)cant(P<0.01, Rayleigh test).