Insect. Soc. 53 (2006) 136?140 0020-1812/06/020136-5 DOI 10.1007/s00040-005-0848-3 ? Birkh?user Verlag, Basel, 2006 Insectes Sociaux Research article Are workers of Atta leafcutter ants capable of reproduction? M.B. Dijkstra and J.J. Boomsma Institute of Biology, Department of Population Biology, University of Copenhagen, Universitetsparken 15, 2100 Copenhagen, Denmark, e-mail: MDijkstra@bi.ku.dk Received 17 January 2005; revised 12 September 2005; accepted 3 October 2005. Summary. Workers of most eusocial Hymenoptera can pro- duce sons after queen loss, which (posthumously) benefi ts the queen and increases worker inclusive fi tness. However, the evolutionary loss of worker ovaries has occurred in sev- eral lineages, while workers in other taxa may be infertile despite having ovaries. Workers of Atta leafcutter ants only lay trophic eggs in queenright colonies. Although Atta colo- nies are commonly kept at universities, museums, and zoos, no reports of worker sons in orphaned colonies exist, suggest- ing that Atta workers are infertile. To explicitly test this, we created eleven orphaned laboratory nests of Atta cephalotes, A. sexdens, and A. colombica, and maintained them for 3?6 months after queen loss. Eight colonies did not produce any brood, but three nests produced 1?4 worker-derived male larvae and pupae. Microsatellite genotyping indicated that these were worker sons. However, all males were tiny (3.5? 9 mm long) compared to normal queen sons (16 mm long), and would almost certainly be unable to mate. We also found reproductive eggs, but most of these had no yolk and were thus inviable. We conclude that Atta workers are not com- pletely infertile, but that worker fertility is low compared to the sister genus Acromyrmex, where workers routinely pro- duce normally-size males after queen loss in the laboratory. We hypothesize that worker reproduction in orphaned Atta fi eld colonies is almost never successful because the last workers die before their sons can be raised to adulthood, but that the importance of worker-laid trophic eggs for queen feeding has precluded the evolutionary loss of worker ova- ries. Keywords: Dwarf males, worker sons, orphaned colonies, worker sterility, worker infertility. Introduction In the majority of eusocial Hymenoptera with physical fe- male castes, workers are unable to mate because they lack genitalia and a functional spermatheca, but typically possess ovaries and can lay unfertilized eggs that develop into males (Choe, 1988; Bourke, 1988). The contribution of the workers to the parentage of the males under queenright conditions (i.e., in the presence of a healthy mother queen) may range from zero to the majority of the males, depending on taxo- nomic group and species (Hammond and Keller, 2003; Wenseleers and Ratnieks, submitted). In species where worker sons are rarely produced in queenright colonies, a major proximate cause of the absence of worker sons is typi- cally the low proportion of reproductive workers (Dijkstra et al., 2005; Wenseleers et al., 2004). The tendency of workers not to form reproductive eggs in the presence of the queen is most frequently explained as an adaptive response to an evo- lutionary history of effi cient queen- or worker policing (Rat- nieks, 1988; Wenseleers et al., 2004a, 2004b). Worker reproduction is always detrimental to the fi tness of the queen as long as she is alive, because the raising of worker sons reduces the amount of colony resources availa- ble to her own sexual offspring. However, the ability of workers to reproduce after the queen?s death is expected to benefi t the deceased queen as well as the workers (Ratnieks, 1988; Franks et al., 1990), and to carry no real costs as the colony is normally doomed. However, obligate worker ste- rility (i.e., the complete absence of ovaries in workers) is reported to occur in at least one genus of stingless bees (Frie- seomelitta: Boleli et al., 1999) and nine genera of ants (Ano- chetus, Eciton, Hypoponera, Leptogenys, Monomorium, Pheidole, Pheidologeton, Solenopsis, Tetramorium: Wheel- er, 1910; Oster and Wilson, 1978; H?lldobler and Wilson, 1990; Villet et al., 1991; but see Whelden, 1963 for Eciton), and it is likely that more examples will be discovered in the future. Insect. Soc. Vol. 53, 2006 Research article 137 Natural selection will only favor obligate worker sterility when four conditions are simultaneously meet: 1. The ontog- eny and/or maintenance of ovaries is costly and reduces the ability of workers to work for the colony; 2. Workers do not use trophic eggs to feed the larvae or the queen; 3. Attempted worker reproduction is never successful in the presence of the queen; 4. Workers do not benefi t from laying reproductive eggs after queen loss, either because the development time of the males is long compared to the life-span of the workers (so that the last workers in the colony are likely to die before any workers sons can be raised to adulthood), or because gynes are only present in the population during a short mating season (so that worker sons from orphaned colonies are unlikely to fi nd a mate). Factors that are expected to further facilitate the evolution of obligate worker sterility are polygyny, a long life-span of queens compared to workers, and the presence of a mechanism for the supersedure of senescent queens and/or ?emergency replacement? of dead queens (cf. Hatch et al., 1999; Faustino et al., 2002), because these reduce the probability that a worker will become orphaned during her life-time. A high degree of queen-worker dimorphism is expected to be another facilitating factor, because this would tend to constrain worker fecundity and thus reduce the pay-off of for worker reproduction. The occurrence of obligate worker sterility in a given spe- cies does not reveal much about the evolution of this trait, as in such a species it is impossible to test how unsuccessful worker reproduction would be after queen loss. A better understanding of the evolution of obligate worker sterility may be gained from species in which workers have retained their ovaries, but never succeed in reproducing after queen loss (hereafter: ?in- fertile workers?). An example of infertile workers is found in the Argentine ant Linepithema humile, where workers can lay trophic eggs but apparently never reproduce in orphaned colo- nies (Aron et al., 2001; S. Aron, pers. comm.). Another candidate taxon for infertile workers are Atta leafcutter ants, which are characterized by a high degree of queen-worker dimorphism and long-lived queens (up to 20 years: Weber, 1972; H?lldobler and Wilson, 1990). While most Atta workers never develop their ovaries in queenright colonies, a minority of the workers have very specialized ovaries which they use to produce trophic eggs to feed to the queen (Dijkstra et al., 2005). Although Atta display colonies are commonly kept at universities, museums, and zoos, to our knowledge only a single dubious report (Tanner, 1892, cited in Wheeler, 1903; see Discussion) of worker sons from orphaned Atta colonies exists. However, this lack of evi- dence is inconclusive, as queenright Atta colonies likewise reproduce only very rarely in the laboratory. The aim of this study was to determine if Atta workers are indeed infertile, as anecdotal evidence suggested (Bazire- B?nazet, 1957; Hernandez, 1998; Dijkstra et al., 2005). We compare our results with those obtained from its sister genus Acromyrmex (Wetterer et al., 1998), where worker sons are routinely produced in orphaned colonies in the laboratory. Identifying the constraints on worker reproduction in Atta may help understand the evolution of obligate worker steril- ity in other taxa. Materials and methods Studying worker reproduction in leafcutter ants Leafcutter ants are obligately dependent on cultivating a clonally propa- gated fungus inside their nest as food. Larval and adult ants eat fungal mycelium, as well as clusters (?staphylae?) of modifi ed hyphal tips (?gongylidia?) (H?lldobler and Wilson, 1990). A problem with studying worker reproduction in Atta is the enormous size (millions of workers and hundreds of fungus gardens) of reproducing colonies in the fi eld. Because our laboratory Atta colonies were far smaller than mature fi eld colonies, our criterion for successful worker reproduction was the pres- ence of worker-derived eggs and fi rst or second instar larvae (hereafter: ?young larvae?), rather than fi nal instar larvae, pupae, or adults. In Acromyrmex leafcutter ants (which have much smaller colonies than Atta), queenright laboratory colonies only produce sexuals when the total volume of the fungus garden(s) is above a certain threshold (Bekkevold and Boomsma, 2000), but orphaned laboratory colonies can contain worker-derived male eggs and young larvae at any fungus vol- ume (Dijkstra, 2005). Orphaned Acromyrmex workers typically raise a large proportion of the worker-derived brood into adult males with nor- mal size and morphology between 14?40 weeks after queen loss, irre- spective of time of year, even if the fungus garden volume is well below the threshold for reproduction by queenright colonies (Dijkstra, 2005). However, orphaned laboratory colonies of Acromyrmex leafcutter ants may also fail to produce any adult worker sons if the fungus garden is extremely small, resulting in the starvation of the larvae in the fi nal in- star. Starving Acromyrmex larvae are not cannibalized by the workers, but dumped intact on the rubbish heap outside the main nest, where they can be easily observed and counted. Maintenance of orphaned nests We divided a total of 10 queenright laboratory colonies of Atta sexdens (n = 2), A. cephalotes (n = 3), and A. colombica (n = 6) into orphaned and queenright nests, by moving 1?2 fungus gardens with a median total volume of 1000 ml (range 800?1900 ml) into separate nest boxes. Before division, the queenright colonies consisted of 1?4 fungus gardens with a median total volume of 2900 ml (range 1750?4000 ml). All colonies had been collected in Panama, except for two A. cephalotes colonies (CE-T1 and CE-T2) from Trinidad. We maintained the orphaned and queenright nests as separate colonies in the dark, at 25 ?C and 70 % RH. Further details of the colony set-up and feeding regime are given in Dijkstra et al. (2005). We regularly inspected the surface of the fungus garden and the rubbish heap for adult sexuals and discarded fi nal instar larvae, re- spectively. Six months after colony division, we killed the orphaned nests by freezing the fungus at ?20 ?C. An exception was the orphaned nest of CE-T2, which was killed three months after colony division. We also froze an additional A. cephalotes nest from Panama (CE-P3), in which the queen had suddenly died four months earlier. We completely dissolved the fungus in 70 % ethanol by gentle stirring. Workers, fungus, and fungus substrate tended to remained suspended, while brood tended to settle on the bottom. We took ca. 200 ml aliquots of the fungus suspen- sion, and carefully searched the bottom layer for any eggs, larvae, or pupae under 6.4x magnifi cation against a dark background. We repeated this procedure until the entire fungus suspension had been searched, which typically took 1?1.5 h per nest. The sex and caste of Atta brood can be easily distinguished from the fi nal larval instar onwards, and their pupae are not enclosed in cocoons. Microsatellite genotyping We found brood in two orphaned A. colombica nests (see Results). For these two nests, we genotyped all larvae and pupae (n = 6) with two microsatellite loci (Etta5-6TF and Etta7-8TF: Fjerdingstad et al., 1998), and reconstructed the genotype of the queen from 30 workers (15 small 138 M. B. Dijkstra and J. J. Boomsma Are Atta workers capable of reproduction? the fungus gardens within the fi rst two months after colony division. We successfully contained these infections by im- mediately removing the affected parts of the fungus. In each orphaned nest of A. sexdens and A. cephalotes, an unusually dense layer of staphylae covered most of the exterior surface of the fungus gardens between 4?6 months after colony divi- sion (Fig. 1), but the density of the staphylae did not differ visibly between orphaned and queenright nests of A. colom- bica. Six months after colony division, the median volume of the orphaned fungus gardens was 600 ml (range 25?1000 ml; or 3?100 % of the original volume), and about 10 % of the workers (median estimate; range <1?25 %) were still alive. One orphaned A. colombica nest (CO-3) contained 30 unembryonated eggs, no larvae, two male pupae that were on the verge of eclosing, but no female (worker or gyne) pupae. These males appeared to have a normal morphology (includ- ing normal-looking genitalia, ocelli, and wings), apart from their small body size (6.5?9 mm long; see Fig. 2a), which was much less than that of males from queenright fi eld colo- nies (?16 mm long; see Fig. 1c). Most eggs (80 %) in CO-3 were completely transparent due to a lack of yolk and were presumably inviable. Another orphaned A. colombica nest (CO-4) contained 44 normal-looking, unembryonated eggs of varying size and shape, two light yellow (ca. 10 d before eclosion) male pupae, one male pre-pupa, one small second instar larva, but no female pupae. The male pupae and pre- pupa were minute (3.5 mm long; Fig. 2b), and had only rudi- mentary wings, but normal-looking genitalia and ocelli. One orphaned A. cephalotes nest (CE-T1) contained a single small second instar larva, but no eggs or pupae of any caste. No eggs, larvae, or pupae of any caste were present in the remaining eight orphaned nests. We never observed dis- carded larvae or pupae on the rubbish heaps of any of the 11 orphaned nests. Genotyping revealed that the queens of CO-3 and CO-4 were both highly multiply mated (me = 5.51 and me = 2.54, respectively), as is typical for leafcutter ants (Villesen et al., 2003), but with slightly higher paternities than reported earlier for Panamanian A. colombica (Fjerdingstad et al., 1998; Mu- rakami et al., 2000). None of the six A. colombica (pre )pupae and larvae were heterozygous for either microsatellite locus, workers and 15 medium workers) per nest. We calculated the effective mating frequency of the queen (me; corrected for binomial sampling er- ror of workers) following formula (2) in Pamilo (1993). We calculated the probability of detecting a worker son (Pd) with formula (2) in Foster et al. (2001). Details about DNA extraction, PCR protocols, and allele scoring are given elsewhere (Fjerdingstad et al., 1998; Hughes et al., 2003). Results The last new workers (derived from queen-laid eggs) pupat- ed in the orphaned nests between 5?8 weeks after colony division, which is 4?16 weeks earlier than in Acromyrmex (Dijkstra, 2005). The primordium of a fungus fruiting body appeared at the top of the fungus garden in one nest (CE-P3) two weeks after the estimated time of death of the queen, but disappeared one month later without having developed fur- ther. In six out of ten orphaned nests (at least one per spe- cies), rotting fungus started to accumulate at the bottom of 5 mm Figure 1. The fungus garden of an orphaned nest of Atta sexdens, with two small workers on top of an unusually dense layer of ?staphylae? (clusters of modifi ed hyphal tips that function as ant food). 1 m m a 1 m m b 1 m m c wg Figure 2. (a) Pupa of a dwarf (6.5 mm) A. colombica worker son from orphaned nest CO-3. (b) Pupa of a dwarf (3.5 mm) A. colombica worker son with reduced wings (marked ?wg?) from orphaned nest CO-4. (c) Normal adult male (16 mm) from a mature colony of A. co- lombica in Panama, with the dwarf males from CO-3 and CO- 4 shown at the same scale. Insect. Soc. Vol. 53, 2006 Research article 139 which is consistent with all being haploid males. Three male pupae (50 %; one and two in CO-3 and CO-4, respectively) were confi rmed worker sons (i.e., were hemizygous for an al- lele not carried by the queen). Corrected for the probability of detecting worker sons (average Pd = 0.57), this is consistent with all six A. colombica males being worker sons. The sex and maternity of the single larva in CE-T1 could not be deter- mined due to a lack of microsatellite loci for A. cephalotes, but its occurrence six months after queen loss strongly sug- gests that it was likewise a worker-derived male. Discussion Production of dwarf worker sons The production of dwarf males in two orphaned nests of A. colombica was surprising. It is highly unlikely that these males would have any mating success because of the small size of their genitalia, which would not match those of gynes, and because of the incomplete wing development in the smallest male pupae. The small size of these worker sons could be due to maternal effects, for example a lower amount of yolk in worker-laid eggs than in queen-laid eggs, or a re- duced attractiveness of worker-derived larvae to nursing workers. Alternatively, their size may have been an artifact of the small size of our laboratory colonies. The latter expla- nation is consistent with the recent fi nding of similar dwarf males (presumably queen sons) in a queenright colony of A. texana in Dallas Zoo (T.R. Schultz, pers. comm.), as well as a report of ?microgynes? produced by queenright laboratory colonies of A. cephalotes (Jutsum and Cherrett, 1978). The production of dwarf sexuals by laboratory colonies seems to be an idiosyncracy of Atta, as Acromyrmex colonies produce normally-sized sexuals in the laboratory (Dijkstra, 2005). Mating opportunities outside the mating season Orphaned workers will only benefi t from producing sons when gynes are present in the population by the time that their sons mature. In the fi eld, queenright Atta and Acro- myrmex colonies typically produce sexuals at the beginning of the rainy season (Weber, 1972). However, during the fi rst 2?25 weeks after queen loss, irrespective of time of year, or- phaned laboratory colonies of Acromyrmex normally produce ?posthumous queen offspring?, i.e. gynes and queen sons that are raised from the remaining queen-derived larvae (Dijkstra, 2005). In the fi eld, these gynes would presumably disperse, and have at least some chance of founding new colonies after mating with males from other orphaned colonies outside the optimal mating season. No ?posthumous queen offspring? were produced by the Atta colonies in this study, nor by a single A. laevigata colony (Hernandez, 1998). However, Wheeler (1903) cited a study by Tanner (1892), who reported the production of gynes and males in an orphaned laboratory colony of A. cephalotes. Wheeler concluded from these ob- servations that Atta workers can parthenogenetically produce females, and this claim has occasionally been cited (e.g., Ledoux 1974; Choe, 1988). However, the reported timing of eclosion (12?15 weeks after queen loss) strongly suggests that these gynes and males were posthumous queen offspring, making this conclusion untenable. Low fertility of Atta workers The presence of worker-derived brood in three colonies (CO-3, CO-4, CE-T1) demonstrates that Atta workers (at least in A. colombica and A. cephalotes) are not completely infertile. However, the high percentage of orphaned nests (72 %) that did not produce any brood, as well as the high proportion of abnormal eggs and the low number of young larvae in the remaining nests, suggest that the fertility of Atta workers is low. This low rate of worker reproduction is par- ticularly striking when compared to its sister genus Acro- myrmex, where a similar sample of orphaned colonies kept under the same laboratory conditions gave at least some nor- mal-sized worker sons in all cases (Dijkstra, 2005). The ob- served accumulation of staphylae in A. cephalotes and A. sexdens indicate that the near-absence of brood was not sim- ply due to a lack of larval food. Given the low proportion of workers that remained alive six months after colony divi- sion, it is unlikely that more brood would have been pro- duced at a later date. In addition, the observed accumulation of rotting fungus in six nests (55 %), and the growth of a fruiting body primordium in one nest, imply that waste dis- posal and fungus tending are ineffi cient after queen loss. Evolution of low worker fertility As explained in the Introduction, obligate worker sterility has evolved in a number of ant genera. Some of these (So- lenopsis, Pheidole, Tetramorium) are very species rich, which suggests that obligate worker sterility has been an ec- ologically successful innovation. We hypothesize that the shorter life-span of Atta workers compared to Acromyrmex implies that orphaned Atta colonies almost never succeed in raising worker sons to adulthood. This constraint will be es- pecially severe under fi eld conditions, where worker survival is predicted to be lower than in the laboratory due to preda- tion, parasitism, and physical wear and tear caused by long- distance foraging. Except for a rare sub-caste of queen nurses with one large ?trophic? ovary, queenright Atta work- ers have embryonic ovaries (Dijkstra et al., 2005), which suggests that the ontogeny of ovaries is costly. We speculate that the importance of worker-laid trophic eggs in the diet of the queen (Bazire-B?nazet, 1970; Dijkstra et al., 2005) has been the only factor that has prevented the evolution of obli- gate worker sterility in Atta, as the high egg-laying rate of their giant queens may critically depend on a continuous supply of trophic eggs. The observed pattern of occasional production of worker sons may therefore represent a non- adaptive byproduct of the selection for the maintenance of ?trophic? worker ovaries. 140 M. B. Dijkstra and J. J. Boomsma Are Atta workers capable of reproduction? Acknowledgements We are grateful to Sylvia Mathiasen for genotyping the males and work- ers, to Ulrich Mueller for collecting the A. cephalotes colonies in Trini- dad, and to David Nash for taking the photo in Figure 1. We are grateful to Ted Schultz and two anonymous referees for their comments on the manuscript. We thank the Smithsonian Tropical Research Institute (STRI) and the Autoridad Nacional del Ambiente (ANAM) for permis- sion to work in Panama, and the Danish Natural Science Research Coun- cil for fi nancial support to J.J. Boomsma. This study was funded by a Ph.D. fellowship from the University of Copenhagen and was done in agreement with the current laws of Denmark. References Aron S., Keller L. and Passera L. 2001. Role of resource availability on sex, caste and reproductive allocation patterns in the Argentine ant Linepithema humile. J. Anim. Ecol. 70: 831?839 Bazire-B?nazet M. 1970. La ponte des ouvri?res d?Atta laevigata Fred. Smith 1858 (Hym. Form.). C. R. Acad. Sc. Paris 270: 1614?1615 Bekkevold D. and Boomsma J.J. 2000. Evolutionary transition to a se- melparous life history in the socially parasitic ant Acromyrmex in- sinuator. J. Evol. Biol. 13: 615?623 Boleli I.C., Paulino-Sim?es Z.L and Bitondi M.M.G. 1999. Cell death in ovarioles causes permanent sterility in Frieseomelitta varia worker bees. J. Morphol. 242: 271?282 Bourke A.F.G. 1988. Worker reproduction in the higher eusocial Hy- menoptera. Q. Rev. Biol. 63: 291?311 Choe J.C. 1988. Worker reproduction and social evolution in ants (Hy- menoptera: Formicidae). In: Advances in Myrmecology (Trager J.C., Ed). E.J. Brill, New York. pp 163?187 Dijkstra M.B., Nash D.R. and Boomsma J.J. 2005. Self-restraint and sterility in workers of Acromyrmex and Atta leafcutter ants. Insect. Soc. 52: 67?76 Dijkstra M.B. 2005. The regulation of male production in leafcutter ants. Ph.D. thesis, Department of Population Biology, University of Co- penhagen, Denmark. x + 111 pp Faustino C.D., Silva-Matos E.V., Mateus S. and Zucchi R. 2002. First record of emergency queen rearing in stingless bees (Hymenoptera, Apinae, Meliponini). Insect. Soc. 49: 111?113 Fjerdingstad E.J., Boomsma J.J. and Thor?n P. 1998. Multiple paternity in the leafcutter ant Atta colombica ? a microsatellite DNA study. Heredity 80: 118?126 Foster K.R., Ratnieks F.L.W., Gyllenstrand N. and Thor?n P.A. 2001. Colony kin structure and male production in Dolichovespula wasps. Mol. Ecol. 10: 1003?1010 Franks N.R., Ireland B. and Bourke A.F.G. 1990. Confl icts, social eco- nomics, and life history strategies in ants. Behav. Ecol. Sociobiol. 27: 175?182 Hammond R.L. and Keller L. 2004. Confl ict over male parentage in so- cial insects. PLoS Biol. 2: e248 Hatch S., Tarpy D.R. and Fletcher D.J.C. 1999. Worker regulation of emergency queen rearing in honey bee colonies and the resultant variation in queen quality. Insect. Soc. 46: 372?377 Hernandez J.V. 1998. Observations of an orphaned colony of Atta laevi- gata (Smith) (Hymenoptera: Attini). Bol. Entomol. Venez. 13: 73?75 H?lldobler B. and Wilson E.O. 1990. The Ants. Springer Verlag, Berlin, Heidelberg. xii + 732 pp Hughes W.H.O., Sumner S., Van Borm S. and Boomsma J.J. 2003. Worker caste polymorphism has a genetic basis in Acromyrmex leaf- cutting ants. Proc. Natl. Acad. Sci. USA 100: 9394?9397 Jutsum A.R. and Cherrett J.M. 1978. Sexuals and a microgyne of Atta cephalotes (L.) (Hym., Formicidae) from laboratory cultures. Ento- mologist?s Mon. Mag. 113: 97 Ledoux A. 1974. Polymorphismus und Kastendetermination bei den Weberameisen. In: Sozialpolymorphismus bei Insekten. Probleme der Kastenbildung in Tierreich (Schmidt G.H., Ed). Wissenschaftli- che Verlagsgesellschaft, Stuttgart. pp 533?542 Murakami T., Higashi S. and Windsor D. 2000. Mating frequency, colo- ny size, polyethism and sex ratio in fungus-growing ants (Attini). Behav. Ecol. Sociobiol. 48: 276?284 Oster G.F. and Wilson E.O. 1978. Caste and Ecology in the Social In- sects. Princeton University Press, Princeton, Guildford. xv + 352 pp Pamilo P. 1993. Polyandry and allele frequency differences between the sexes in the ant Formica aquilonia. Heredity 70: 472?480 Villesen P. and Boomsma J.J. 2003. Patterns of male parentage in the fungus-growing ants. Behav. Ecol. Sociobiol. 53: 246?253 Ratnieks F.L.W. 1988. Reproductive harmony via mutual policing by workers in eusocial Hymenoptera. Am. Nat. 132: 217?236 Villesen P., Murakami T., Schultz T.R. and Boomsma J.J. 2002. Identify- ing the transition between single and multiple mating of queens in fungus-growing ants. Proc. R. Soc. Lond. B. 269: 1541?1548 Villet M.H., Crewe R.M. and Duncan F.D. 1991. Evolutionary trends in the reproductive biology of ponerine ants (Hymenoptera: Formici- dae). J. Nat. Hist. 25: 1603?1610 Weber N.A. 1972. Gardening Ants: the Attines. American Philosophical Society, Philadelphia. xvii + 146 pp Wenseleers T., Hart A.G. and Ratnieks F.L.W. 2004. When resistance is useless: policing and the evolution of reproductive acquiescence in insect societies. Am. Nat. 164: 154?167 Wenseleers T., Helanter? H., Hart A., and Ratnieks F.L.W. 2004a. Worker reproduction and policing in insect societies: an ESS analy- sis. J. Evol. Biol. 17: 1035?1047 Wetterer J.K., Schultz T.R. and Meier R. 1998. Phylogeny of the fungus- growing ants (Tribe Attini) based on mtDNA sequence and mor- phology. Mol. Phyl. Evol. 9: 42?47 Wheeler W.M. 1903. The origin of female and worker eggs of partheno- genetic workers. Science 18: 830?833 Wheeler, W.M., 1910. Ants. Their Structure, Development and Behavior. Columbia University Press, New York. xxv + 663 pp Whelden R.M. 1963. Anatomy of adult queen and workers of army ants Eciton burchelli Westw. and E. hamatum Fabr. (Hymenoptera: For- micidae). J. N. Y. Entomol. Soc. 71: 14?30, 90?115, 158?178, 246? 261