Naturwissenschaften 85, 133–135 (1998) © Springer-Verlag 1998 that colony propagation is achieved solely through these queens. Never- theless our laboratory experiments de- Are Workers Capable monstrate that workers are capable of of Colony Foundation? raising new colonies on their own. We excavated 75 H. saltator colonies Ju¨rgen Liebig, Bert Ho¨lldobler in the western Ghats in southern India between 1991 and 1995. Twelve of Theodor-Boveri-Institut, LS Verhaltensphysiologie und Soziobiologie, these were used as stock colonies for Am Hubland, D-97074 Wu¨rzburg, Germany the following experiments: Male co- Christian Peeters1 coons had been added several weeks earlier, because H. saltator workers CNRS URA 667, Laboratoire d’ Ethologie expe´rimentale et compare´e, normally inbreed with males that Universite´ Paris Nord, F-93439 Villetaneuse, France eclose in their own nests (Peeters et al. 1998). Large numbers of young workers mate annually, but most of these are inhibited from producing Received: 23 September 1997 / Accepted in revised form: 5 January 1998 eggs by the existing reproductives. To obtain the infertile workers required for our experimental groups we se- In the ponerine ant lish new colonies, and then concen- lected workers inside the nest cham- saltator both female castes reproduce trate exclusively on reproduction bers of the stock colonies because sexually, but only queens disperse and (Ho¨lldobler and Wilson 1990). these are likely to be young and thus independently start new colonies. Queens can found colonies either with inseminated. To obtain newly differ- Although incipient worker founda- or without the help of sterile workers, entiated gamergates we removed es- tions have never been found in the i.e., dependently or independently tablished gamergates from the various field, mated workers of H. saltator, sin- (Ho¨lldobler and Wilson 1977). In a stock colonies, and after 2–3 weeks gly or in groups of three, proved cap- number of species of the subfamily new reproductives had started to lay able of producing and raising adult off- , however, a queen caste eggs and engaged in aggressive inter- spring in the laboratory. Of 25 worker does not exist, and reproduction is actions, which made them clearly re- foundations 16 produced adult work- carried out by mated, egg-laying cognizable. To investigate the work- ers after 160 days, and two groups pro- workers, which are called gamergates ers’ abilities we isolated them in pre- duced 110 and 167 workers after 1 (Peeters and Crozier 1988; Peeters formed nests (plastic boxes year. Any mated worker has the option 1993). Studies have shown that such (19×9 cm) with a floor of plaster of of remaining in the natal nest to rear gamergate colonies propagate by fis- Paris containing two circular cham- closely related brood or of leaving to sion, a dependent strategy of colony bers (4.5 cm diameter) which were found a new colony. Reasons for the foundation. Except for a few anecdo- covered with a glass plate) and dis- absence of worker foundations in na- tal reports (Haskins and Whelden tributed them in the following experi- ture can be attributed to differences in 1965; Ward 1981) it is not known mental groups: I, 10 single infertile morphological and physiological spe- whether gamergates are also able to workers; II, 5 groups each containing cialization between the two female establish new colonies independently. 3 infertile workers; III, 5 single ga- castes. Higher body reserves of queens We investigated in the laboratory mergates; IV, 5 groups each contain- compared to workers may be the essen- whether workers of the ponerine ant ing 3 gamergates. We also collected tial factor for the absence of worker H. saltator retain the behavioral reper- four dealate queens walking on the foundations in because they re- tory for independent colony founda- ground in India, and one queen from duce the need to forage outside, and tion, although this behavior may not an incipient nest in which she had al- thus predation and parasitization risk necessarily be expressed. ready laid eggs. We followed the and buffer periods of low food avail- In H. saltator both queens and gamer- course of colony foundation by these ability during foundation. gates reproduce (Peeters and Ho¨lldob- queens in the laboratory to obtain In the two female castes differ in ler 1995). The size dimorphism be- baseline data for assessing the founda- both morphology and behavior. While tween queens and workers is small tion success of workers. sterile workers generally help to rear but distinct. Queens disperse by flight Most of the worker and queen groups new offspring, queens disperse, estab- and found colonies alone. When the succeeded in producing and raising queen dies in the course of colony on- adult offspring; however, despite ad togeny several gamergates take over libitum supply of live crickets which 1 Present address: CNRS URA 258, Labora- reproduction, but the colonies con- they hunt, 20–40% of the workers in toire d’ Ecologie, Universite´ Pierre-et-Marie tinue to produce winged queens and the experimental groups died before Curie, 7 quai Saint Bernard, F-75252 Paris males every year (Peeters and Ho¨ll- they produced adult offspring (Ta- CEDEX 05, France dobler 1995; Peeters et al. 1998). ble 1). The successful worker founda- Correspondence to: J. Liebig Based on our field data we suggest tions (n=16/25) produced between 2

Naturwissenschaften 85 (1998) © Springer-Verlag 1998 133 Table 1. Mortality in each experimental group and founding success after 160 days

Initiated by Infertile workers Gamergates Single queen

Group size 1 3 1 3 1

Number of groups 10 5 5 5 5 Mortality within 90 daysa 2 (20%) 5 (33%) 2 (40%) 5 (33%) 1 (20%) Unsuccessful foundation 4 b 13c 11 (no adult workers produced)

aNo mortality occurred between 90 and 160 days b Of the two surviving workers one produced only eggs within 160 days, and the other pro- duced only two males c One worker produced only males, thus she was probably not a gamergate

the speed of colony foundation. Two workers have never been found in the groups revealed the high reproductive field (Ward 1981). potential of workers: a single worker Reasons for the absence of an alter- Fig. 1. Success of experimental foundations which was originally infertile pro- native independent founding strategy based on the number of cocoons and adult duced two males within the first 23 by single workers or small worker workers after 90 and 160 days of isolation weeks, suggesting that it was not ma- groups become evident when consid- (data on mortality can be found in Table 1). ted. Another 13 weeks later, however, ering the risks involved. In the natural Initially sterile workers are indicated as work- the first worker eclosed, indicating situation foundresses are confronted ers. In the single gamergate foundations that the founding worker had mated with the risks of predation and mean values are not given due to the small sample size. Worker and cocoon production with its son. In a second case a single parasitism, which can be substantial were not significantly different between the gamergate had produced 6 workers in foraging workers (Porter and different foundations after 90 days (Kruskal- before she died. These virgin workers Jorgensen 1981; Schmid-Hempel and Wallis test for workers, ANOVA for co- produced males, and after several Schmid-Hempel 1984; Ho¨lldobler and coons). After 160 days only cocoon produc- weeks new workers eclosed, indicat- Wilson 1990). Queens of the “higher” tion between single queens and single work- ing again that sons can mate with ants escape these problems by found- ers differed significantly (P<0.05, ANOVA their mothers. ing claustrally. They are able to rear with Tukey’s test for unequal n) The laboratory conditions in which their first offspring without foraging workers successfully produced and outside the nest chamber since they and 14 workers within 160 days raised adult offspring were very artifi- metabolize their body reserves. In- (Fig. 1), and continued to grow. The cial because foundations were pro- deed, the fat content of queens before results of 12 additional experiments vided with a secure nest chamber and mating is highest in those species ex- were very similar to those described foundresses hunted on crickets sup- hibiting claustral founding (Keller and in Fig. 1, but these were not pooled plied ad libitum. Thus we think that Passera 1989). Additionally, queens with the other data since the workers these results cannot be extended to accumulate storage proteins before had been collected from the same the natural situation. Indeed, we never leaving their parental nest to found a stock colonies, i.e., the data were not collected worker foundations in the new colony (Wheeler and Buck statistically independent. Two found- field. In 75 excavated colonies smaller 1995). They also histolyze their flight ing groups that consisted of one and colonies always contained a queen muscles to feed the first generation of three originally infertile workers, (Liebig et al. 1998), and colony found- workers (Janet 1907), and, accord- reached sizes of 110 and 167 workers, ing by budding was never observed in ingly, the size dimorphism between respectively, after 1 year. In the stan- the laboratory or in the field. queens and workers is greatest in spe- dardized experiments only cocoon The ability of single ant workers to cies that establish new colonies claus- production in single worker and sin- rear offspring in the laboratory has trally (Stille 1996). gle queen foundations differed signifi- only been recorded anecdotally twice Queen-worker dimorphism is gener- cantly after 160 days of isolation in the literature. Isolated virgin work- ally small in ponerine ants (Peeters (P<0.05, analysis of variance with Tu- ers of metallica reared 1993; Liebig et al. 1995), and the key’s test for unequal n), but the fol- males, suggesting that isolated mated queens probably lack storage proteins lowing trend was apparent: productiv- workers are similarly capable of pro- (Wheeler and Martinez 1995). In- ity was lowest in single worker foun- ducing workers (Haskins and Whel- dependent foundation is generally dations, intermediate in groups of den 1965). In R. chalybaea an iso- semiclaustral, and queens regularly three workers, and highest in queen lated gamergate produced three work- leave the nest chamber for foraging foundations (Fig. 1). ers within 5 months, which was the trips. H. saltator queens weigh The initial developmental state of the same as that reared by a conspecific 21.1±2.1 mg (n=12) and workers ovaries (infertile workers versus ga- queen during the same period. How- 12.3±2.3 mg (n=28), i.e., queens are mergates) apparently did not affect ever, founding groups of one to three 1.7 times heavier than workers. Since

134 Naturwissenschaften 85 (1998) © Springer-Verlag 1998 flight muscle histolysis can provide designed to protect against enemies Keller L, Passera L (1989) Size and fat-con- additional body reserves (Haskins and flooding in an area with heavy tent of gynes in relation to the mode of 1970), H. saltator queens are physi- seasonal rains (Peeters et al. 1994), colony founding in ants (, they can suffer accidental destruction. Formicidae) Oecologia 80:236–240 cally better specialized to found colo- Liebig J, Heinze J, Ho¨lldobler B (1995) nies semiclaustrally than workers. In While this may be a relatively rare Queen size variation in the ponerine ant addition to a reduction in foraging event, workers are capable of continu- Ponera coarctata (Hymenoptera: Formici- time, which decreases mortality risk, ing the colony or starting a new one, dae). Psyche 102:1–12 greater body reserves help to buffer either alone or with other nestmates Liebig J, et al (1998) (in preparation) periods of low or fluctuating foraging and brood. Alternatively, the retention Maschwitz U, Hahn M, Schonegge P (1979) success. The significance of such fluc- of the ancestral trait of independent Paralysis of prey in ponerine ants Harpeg- tuations for semiclaustrally founding colony foundation by workers can be nathos saltator and chinensis. Naturwissenschaften 66:213–214 queens can be deduced from a com- explained simply by evolutionary in- Peeters C (1993) Monogyny and polygyny in parison with solitary hymenopterans ertia since they have evolved from ponerine ants with or without queens. In: where fluctuations in food availability species with totipotent females. Keller L (ed) Queen number and affect the number and size of off- More generally, workers of H. salta- in insects. Oxford University Press, New spring (Rosenheim et al. 1996; tor have fitness advantages when York Strohm and Linsenmair 1997). The staying in their parental colony. Infer- Peeters C, Crozier RH (1988) Caste and re- problem of fluctuating foraging suc- tile workers gain by helping highly production in ants: not all mated egg- layers are “queens.” Psyche 95:283–288 cess is best illustrated in the central related nestmates, but there is also the Peeters C, Ho¨lldobler B (1995) Reproductive limit theorem (Wenzel and Pickering chance to become a gamergate in a cooperation between queens and their 1991), which posits that fluctuations secure nest in which their relatives mated workers: The complex life history in the daily food intake of a colony have already invested a lot. Although of an ant with a valuable nest. Proc Natl increases with decreasing colony size, under laboratory conditions H. salta- Acad Sci USA 92:10977–10979 reaching a maximum in single-queen tor workers show a remarkable behav- Peeters C, Ho¨lldobler B, Moffett M, foundations. ioral plasticity in their ability to found Musthak-Ali TM (1994) Wall-papering and colonies independently, under natural elaborate nest architecture in the ponerine In addition to drawing on their body ant . Ins Soc reserves, H. saltator foundresses can conditions they are clearly disadvan- 41:211–218 buffer episodes of temporarily low taged as colony foundresses compared Peeters C, et al (1998) (in preparation) foraging success by eating already to the queen castes. Thus the only Porter SD, Jorgensen CD (1981) Foragers of present eggs or larvae which, how- successful strategy of workers seems the harvester ant, Pogonomyrmex owyheei: ever, increases the period until the to stay in a colony as hopeful future a disposable caste? Behav Ecol Sociobiol eclosion of the first workers. In addi- reproductives, although most of them 9:247–256 tion, their ability to store paralyzed remain sterile helpers. Rosenheim JA, Nonacs P, Mangel M (1996) Sex ratio and multifaceted parental invest- prey (Maschwitz et al. 1979) reduces ment. Am Nat 148:501–535 the number of foraging trips because We are grateful to Raghavendra Schmid-Hempel P, Schmid-Hempel R (1984) larvae and foundress can use a single Gadagkar for his hospitality and logis- Life duration and turnover of foragers in piece of prey for several days. Never- tic support during several visits in the ant Cataglyphis bicolor (Hymenoptera, theless, as soon as the prey has been India. We thank two anonymous refer- Formicidae) Ins Soc 31:345–360 eaten up, new foraging trips become ees for helpful comments on the Stille M (1996) Queen/worker thorax volume necessary, and the foundress again manuscript. This study was supported ratios and nest-founding strategies in ants. Oecologia 105:87–93 faces fluctuating foraging success. by the DAAD (PROCOPE program) Strohm E, Linsenmair KE (1997) Low re- Thus, trophic unpredictability is a and DFG SFB 251. source availability causes extremely male- further factor that may have led to the biased investment ratios in the European evolution of claustral foundation in beewolf, Philanthus triangulum F (Hyme- ants in general. Haskins CP (1970) Researches in the biology noptera, Sphecidae). Proc R Soc Lond B Another general advantage of queen and social behavior of primitive ants. In: 264:423–429 foundation over worker foundation is Aronson LR, Tobach E, Lehrman DS, Ro- Ward PS (1981) Ecology and life history of senblatt JS (eds) Development and evolu- the Rhytidoponera impressa group (Hyme- aerial dispersal by queens. Workers noptera: Formicidae). II. Colony origin, cannot colonize new localities or re- tion of behavior. Freeman, San Francisco Haskins CP, Whelden RM (1965) “Queen- seasonal cycles, and reproduction. Psyche colonize areas in a fragmented habitat lessness”, worker sibship, and colony ver- 88:109–126 where the former population became sus population structure in the formicid Wenzel JW, Pickering J (1991) Cooperative extinct. Every successful worker foun- genus Rhytidoponera. Psyche 72:87–112 foraging, productivity, and the central limit dation would be close to her parental Ho¨lldobler B, Wilson EO (1977) The number theorem. Proc Natl Acad Sci USA 88:36– colony, where she would eventually of queens: an important trait in ant evolu- 38 compete for local resources. tion. Naturwissenschaften 64:8–15 Wheeler DE, Buck NA (1995) Storage pro- teins in ants during development and colo- The only conceivable scenario in Ho¨lldobler B, Wilson EO (1990) . Belknap, Cambridge ny founding. J Ins Physiol 41:885–894 which the founding behavior of work- Janet C (1907) Anatomie du corselet et histo- Wheeler DE, Martinez T (1995) Storage pro- ers could be expressed is the partial lyse des muscles vibrateurs, apre`s le vol teins in ants (Hymenoptera, Formicidae) destruction of the parental colony. nuptial chez la reine de la fourmi (Lasius Comp Biochem Physiol B 112:15–19 Although the nests of H. saltator are niger). Ducourtieux et Gout, Limoges

Naturwissenschaften 85 (1998) © Springer-Verlag 1998 135