THE ROYAL SOCIETY FirstCite? e-publishing Received 20 February 2002 Accepted 16 April 2002 Published online Identifying the transition between single and multiple mating of queens in fungus-growing ants Palle Villesen1, Takahiro Murakami2, Ted R. Schultz3 and Jacobus J. Boomsma1'4* 1 Department of Ecology and Genetics, University of Aarhus, DK-8000 Aarhus C, Denmark 2Section of Integrative Biology, Patterson Laboratories, University of Texas, Austin, TX 78712, USA ^Department of Systematic Biology, MRC 188, National Museum of Natural History, Smithsonian Institution, IPWMMgWM, DC 20073-7072, U&4 4Zoological Institute, Department of Population Ecology, University of Copenhagen, Universitetsparken 15, DK 2100 Copenhagen, Denmark Obligate mating of females (queens) with multiple males has evolved only rarely in social Hymenoptera (ants, social bees, social wasps) and for reasons that are fundamentally different from those underlying multiple mating in other animals. The monophyletic tribe of ('attine') fungus-growing ants is known to include evolutionary derived genera with obligate multiple mating (the Acromyrmex and Atta leafcutter ants) as well as phylogenetically basal genera with exclusively single mating (e.g. Apterostigma, Cyphomyr- mex, Myrmicocrypta). All attine genera share the unique characteristic of obligate dependence on symbiotic fungus gardens for food, but the sophistication of this symbiosis differs considerably across genera. The lower attine genera generally have small, short-lived colonies and relatively non-specialized fungal sym- bionts (capable of living independently of their ant hosts), whereas the four evolutionary derived higher attine genera have highly specialized, long-term clonal symbionts. In this paper, we investigate whether the transition from single to multiple mating occurred relatively recently in the evolution of the attine ants, in conjunction with the novel herbivorous 'leafcutter' niche acquired by the common ancestor of Acromyrmex and Atta, or earlier, at the transition to rearing specialized long-term clonal fungi in the common ancestor of the larger group of higher attines that also includes the genera Trachymyrmex and Sericomyrmex. We use DNA microsatellite analysis to provide unambiguous evidence for a single, late and abrupt evolutionary transition from exclusively single to obligatory multiple mating. This transition is historically correlated with other evolutionary innovations, including the extensive use of fresh vegetation as substrate for the fungus garden, a massive increase in mature colony size and morphological differen- tiation of the worker caste. Keywords: fungus-growing ants; microsatellite markers; multiple mating; paternity; relatedness; attine ants 1. INTRODUCTION Multiple mating of queens in social Hymenoptera is a derived trait, occurring only in relatively few evolutionary derived taxa of ants, social bees and social wasps (Boomsma & Ratnieks 1996). A possible reason for this remarkable pattern is that kin selection on helping behav- iour is less effective when workers are mostly half sisters than when they are full sisters (Hamilton 1964). Multiple queen mating is therefore only evolutionary stable after helpers (workers) have irreversibly lost their ability to mate and when the behaviour significantly enhances queen fit- ness. The possible fitness benefits to queens of social Hymenoptera are fundamentally different from those applying to females of non-social animals (Boomsma & Ratnieks 1996; Strassmann 2001), because queens of social insects choose the prospective father(s) of their off- spring only once and do not subsequently re-mate. Cur- rent hypotheses to explain multiple mating in social insects have therefore concentrated on sperm limitation and on "Author for correspondence (jjboomsma@zi.ku.dk). Proc. R. Soc. Lond. B DOI 10.1098/rspb.2002.2044 the advantages of genetically diverse offspring, rather than on female choice of superior mates and sperm displace- ment (Boomsma & Ratnieks 1996; Schmid-Hempel 1998; Strassmann 2001). Multiple mating in social insects is thus predicted to be associated with high longevity and fecundity of queens, traits that are typical for the evolu- tionary more advanced species of ants, social bees and social wasps, and, indeed, it is in some of these taxa that obligate multiple mating has been found (Boomsma & Ratnieks 1996). The 12 genera of fungus-growing ants (Formicidae: Attini) contain ca. 210 described species (Chapela et al. 1994; Schultz & Meier 1995; Wetterer et al. 1998), all of which obligately cultivate mutualistic fungi upon which they are completely dependent (Weber 1972). Based on their fungal associations, attine ants can be divided into two distinct groups. On the one hand the lower attines are characterized by small colony sizes, a relatively hetero- geneous collection of symbionts with free-living counter- parts, and rather generalized behavioural interactions with these fungi (Chapela et al. 1994; Mueller et al. 1998). Higher attines, on the other hand, cultivate a highly derived, monophyletic group of fungi lacking known 02PB0145.1 ? 2002 The Royal Society 02PB0145.2 P. Villesen and others The transition between single and multiple mating in attine ants free-living relatives and possess unique adaptations for interacting with these fungi (Chapela et al. 1994; Mueller et al. 2001). Within the four genera of higher attine ants, there is another major, but more gradual evolutionary transition. The basal genera Trachymyrmex and Sericomyr- mex have approximately monomorphic worker castes, as in the lower attines. Colony sizes of Trachymyrmex spp. are hardly larger than the largest colonies of lower attines and they exploit green leaves and flowers only marginally better than do the lower attines. However, the colonies of some Sericomyrmex spp. may be an order of magnitude larger. A significant part of the diet of Sericomyrmex spp. consists of flowers and fruit, but they have not been observed to process green leaves to any significant degree in the field. The two most derived genera Acromyrmex and Atta are often referred to as the leafcutter ants. They have large to very large colonies, rely exclusively on fresh veg- etation (mostly leaves), and have highly polymorphic worker castes (Wilson 1971; Weber 1972; Holldobler & Wilson 1990). Previous studies with microsatellite DNA markers have shown that lower attines have singly mated queens (Villesen et al. 1999; Murakami et al. 2000), whereas the most derived higher attines, the Atta and Acromyrmex leaf- cutting ants, are invariably multiply mated (Fjerdingstad et al. 1998; Bekkevold et al. 1999; Boomsma et al. 1999; Fjerdingstad & Boomsma 2000). Because queen mating frequencies of leafcutter ants are the highest so far docu- mented for ants, these comparative data imply that funda- mental changes in selection on the mating system have taken place during the evolutionary history of the attine ants. The objectives of the present study are: (i) to assess whether a single transition or multiple transitions from single to multiple queen mating occurred; (ii) to assess whether such transition(s) occurred in the common ances- tor of all higher attine ants and/or in the common ancestor of the leafcutter ants, i.e. to assess whether they occurred in conjunction with the shift to rearing specialized long- term clonal fungi, or in conjunction with entering the her- bivorous 'leafcutter' niche; (iii) to provide a first estimate of the distinctness of the transition(s) by inferring the like- lihood that facultative multiple mating occurs in at least some of the extant genera; (iv) to re-evaluate the earlier data on queen mating frequency in attine ants with novel statistical methods, allowing the estimation of confidence limits around the means obtained; and (v) to make a more precise inference of the selection forces that may have pro- moted the evolution and maintenance of multiple queen mating. To fulfil these objectives, we analysed queen mating-frequency in five species of the key genera Trachy- myrmex and Sericomyrmex, by using a series of highly vari- able DNA microsatellite markers (Villesen et al. 1999, 2002). (a) Natural history and systematic position of Trachymyrmex and Sericomyrmex The genus Trachymyrmex currently consists of 41 described species. These are geographically widespread in the Americas, ranging from the northern United States to Argentina. Worker size is strictly monomorphic in some species and slightly polymorphic in others (Holldobler & Wilson 1990; Beshers & Traniello 1994). Trachymyrmex spp. seem to prefer insect faeces, dead vegetable matter and flower parts as primary substrates for their fungus gardens. Worker number in mature colonies of the investi- gated species varies from ca. 100-500 (Weber 1966, 1972; Holldobler & Wilson 1990; J. J. Boomsma, personal observation), which implies that colony sizes are generally larger than those typical for the lower attines but orders of magnitude smaller than those of the leafcutter ants. Based on available data, Trachymyrmex colonies have a single queen (Murakami 1998; J. J. Boomsma, unpub- lished data). The genus Sericomyrmex currently consists of 20 described species, which are easily distinguished from other attines by the abundant flexuous hairs on their cuti- cle. The genus is distributed from Mexico to southern Brazil. In the field, Sericomyrmex spp. use fruit, flowers, some leaves and possibly dead vegetable matter as fungal substrates (Urich 1895; Weber 1945, 1972). Experimental laboratory colonies readily accept, cut and use fresh leaves of bramble (Rubus sp.; not occurring in the native range of the ants), showing that these ants can handle fresh leaves when available in or very close to the nest (P. Vill- esen, personal observation). Colony sizes range from less than 100 to ca. 3000 workers (Weber 1972; Murakami 1998; T. R. Schultz, personal observation). The mature colonies of the Sericomyrmex species investigated here had a few thousand workers and several garden chambers. Workers were monomorphic, but a few species with a slight polymorphism in the worker caste are known (Holldobler & Wilson 1990). A recent study by Murakami et al. (2000), based on a CA-repeated primed-PCR DNA fingerprinting technique, has suggested that some Trachy- myrmex species are singly mated, but that multiple mating occurs in Sericomyrmex amabilis. This technique allows the estimation of colony kin-structure based on band-sharing statistics, but does not produce accurate assessments of the effective queen mating frequency and of possible con- founding effects of facultative polygyny or occasional unrelated workers due to raids on or merging with neigh- bouring colonies (Adams et al. 2000). As multiple queens are occasionally found in Panamanian Sericomyrmex colon- ies (J. J. Boomsma, personal observation), we therefore reanalysed the S. amabilis samples of Murakami et al. (2000) with our DNA microsatellite markers in order to distinguish between multiple queen mating and polygyny. The higher attines as a whole {Trachymyrmex, Sericomyr- mex, Acromyrmex and Atta) are a monophyletic group (Chapela et al. 1994; Schultz & Meier 1995; Wetterer et al. 1998). Within the higher attines, the leafcutting genera Atta and Acromyrmex comprise a derived, monophyletic group, whereas Sericomyrmex is the basal monophyletic clade of the higher attines (Schultz & Meier 1995; Schultz 2000). The genus Trachymyrmex as currently defined is probably paraphyletic, both with respect to Sericomyrmex and with respect to {Acromyrmex plus Atta). Thus, some Trachymyrmex species will ultimately have to be trans- ferred to the genus Sericomyrmex, whereas one or a few others (e.g. T. diversus, which does not occur in Panama) may be transferred to the genus Acromyrmex (Schultz & Meier 1995; Schultz 2000; T. R. Schultz, unpublished data). Natural history data for Acromyrmex and Atta have been summarized by various authors (e.g. Weber 1972, 1979; Fowler et al. 1986; Holldobler & Wilson 1990). Proc. R. Soc. Lond. B The transition between single and multiple mating in attine ants P. Villesen and others 02PB0145.3 2. METHODS (a) Population sampling Field collections were made in and around Gamboa, Fort Sherman and El Llano in the Republic of Panama. Colonies of Trachymyrmex cf. zeteki (n = 8), T cometzi sp. 1 (re = 6) and T cometzi sp. 2 in = 10) were collected in May-June 1998, whereas colonies of S. amabilis in = 5) and 5. cf. amabilis (re = 5) were collected in spring 1996 and May-June 1998. The colonies of 5. amabilis (re = 4) previously analysed by Murakami et al. (2000) were collected in 1996-1997 at Barro Colorado Island, ca. 18 km from Gamboa. Mother queens and offspring workers were genotyped on an ALF express automatic sequencer for three to four microsatellite loci per species (see also below). The genera Trachymyrmex and Sericomyrmex have not been revised taxonomically, and many new and cryptic species are likely to exist. The names used in this study are the most accur- ate that are currently available. In spite of the lack of described species, the molecular data indicate that all species identified are proper taxonomic entities with homogeneous non-overlapping gene pools for the marker loci. Voucher specimens have been deposited at the National Museum of Natural History (USNM), Smithsonian Institution, Washington, DC, under species names used in the present paper: 5. amabilis, S. cf. amabilis, T cf. zeteki, T. cometzi sp. 1 and T. cometzi sp. 2. (b) DNA extractions and PCR DNA for genotyping individual ants was extracted using a modified Chelex 100 extraction protocol (Walsh et al. 1991). After homogenization, samples were incubated for 1 h at 55 ?C in 200 \iX 5% Chelex solution (5% Chelex in 50 mM Tris pH 8.0), followed by vortexing for 30 s and boiling for 12 min. The DNA was then centrifuged at 13 000 rpm for 2 min and stored at 20 ?C until use. PCR reactions were carried out in 6 (0.1 volumes containing 1 (0.1 of extracted template DNA, 2 pmol of each PCR primer (one primer was 5'Cy end-labelled for specific use on the ALF express), 0.2 mM dATP, dCTP, dGTP and dTTP, one unit of Tag polymerase buffer and 0.375 units of Tag polymerase (Pharmacia). After testing a total of 32 microsatellite loci (Fjerdingstad et al. 1998; Villesen et al. 1999, 2002; Ortius-Lechner et al. 2000) for amplification and variability, we decided to use six of these loci in the present analysis of queen mating frequency in Trachy- myrmex and Sericomyrmex. The loci used had 4.9 ? 2.4 alleles, with expected heterozygosities of 0.697 ?0.159 (mean ? s.d.). Two of the markers (Cypho9_10, Cyphol5B_16B) were orig- inally developed for Cyphomyrmex longiscapus (Villesen et al. 2002) and one of these, Cyphol5B_16B, was highly variable in all five species. We further used three markers (Trachy 3_4, Tra- chy 5_6, Trachy 11_12) developed for T. cf. zeteki (Villesen et al. 2002) and one locus (Ech3385) developed for Acromyrmex echinatior (Ortius-Lechner et al. 2000). (c) Phytogeny Bootstrap values are based on 100 bootstrap pseudoreplicates (Felsenstein 1985) obtained with maximum-likelihood phylo- genetic analyses of an 804 bp fragment of the nuclear EF-la gene and an 807 bp fragment of the mitochondrial COI gene using the computer program PAUP 4.0b4a (Swofford 2001). Likelihood analyses employed the most complex model available (general time-reversible with gamma-distributed rates and a pro- portion of nucleotide sites assumed to be invariant; Rodriguez et al. 1990). Due to the model and methods employed and the bootstrap values obtained (all greater than or equal to 85%), the accuracy of the topology relating the higher attine genera and species is expected to be robust at greater than 95% confidence (Hillis & Bull 1993). (d) Data analysis The number of haploid fathers per colony could easily be derived by comparing the mother-offspring genotypes. We nor- mally genotyped the mother queen and 10-15 workers (12.89 ?2.98) from each colony with three microsatellite loci. In one case (T. cf. zeteki), a fourth locus was applied to see how it affected the quantitative analysis and conclusions. The results showed no difference in the accuracy of effective paternity esti- mates between the four-locus analysis in this species and the three-locus analyses in the four other species. Two sources of error are possible in the estimation of queen mating frequencies from genotypic pedigree data. Both a non- detection error, due to limited genetic variation at marker loci, and an error of non-sampling of paternal genotypes when the number of offspring analysed is relatively small, may lead to an underestimation of queen mating frequency (Boomsma & Rat- nieks 1996). Correction procedures for both sources of error exist (Pamilo 1982, 1993; Boomsma & Ratnieks 1996; Ped- ersen & Boomsma 1999) and we have followed the procedure of Pedersen & Boomsma (1999) by using a beta version of the MATESOFT program (Moilanen & Pedersen 2001). In some col- onies (re = 9) the queen was not found, so her genotype was inferred by the computer program (Moilanen & Pedersen 2001), which takes limited sample size and limited resolution of genetic markers into account. If more than one combination of queen- mate (s) is possible, the program assigns probabilities to different combinations and uses these in the subsequent estimations of errors and corrected frequencies of possible double matings (see table 1 for details). We calculated the 95% confidence intervals (CI) of the observed fraction of multiply mated queens F (where F can be equal to zero) according to a method by Schmid-Hempel & Schmid-Hempel (2000). The upper 95% CI of an estimate in which no cases of multiple mating were observed (Dobs = 0) is 1 I)' where n is the number of colonies analysed and a is the desired significance level (Zar 1999). After correcting for the sampling and detection errors estimated by the MATESOFT program (Moilanen & Pedersen 2001), the 95% CI is obtained as AXP 0; I"/ where / is the combined detection and sampling error according to Pedersen & Boomsma (1999). As the observed number of multiple matings in the present dataset is zero, we had to use a hypothetical value for paternity skew. As an estimate, we have used the average uncorrected value of 19 ant species, c = 0.77 (corrected value c = 0.753) (Boomsma & Ratnieks 1996). Higher hypothetical values of c will result in larger errors and thus higher estimates of Dcst, the inferred fraction of multiply mated queens in the population. However, high values of paternity skew make inferred cases of multiple mating ineffective for increasing genetic variation among offspring, so that relatedness estimates are hardly affected by different values of Proc. R. Soc. Lond. B 02PB0145.4 P. Villesen and others The transition between single and multiple mating in attine ants Table 1. Paternity and offspring relatedness in five species of Trachymyrmex and Sericomyrmex fungus-growing ants. Trachymyrmex Trachymyrmex Trachymyrmex Sericomyrmex Sericomyrmex cf. zeteki cometzi sp. 1 cometzi sp. 2 amabilis cf. amabilis observed frequency of double matings (Dobs) number of microsatellite loci used number of colonies analysed number of offspring workers genotyped (mean + s.d.) upper 95% CI for frequency of double queen mating (uncorrected) combined error for identification of double mating (/)' upper 95% CI for frequency of double queen mating (corrected)"3 maximal effective paternity of nest-mate workers' minimal inferred relatedness among nest-mate workers'* 0 0 0 0 0 4 3 3 3 3 8 6 10 11 5 13.0 + 4.6 12.0 + 0.0 13.5 + 3.7 10.9 + 3.8 12.2+1.8 0.369 0.459 0.308 0.285 0.522 0.075 0.066 0.090 0.107 0.059 0.399 0.492 0.339 0.319 0.555 1.165 1.211 1.136 1.127 1.244 0.679 0.663 0.690 0.694 0.652 a Estimated by the MATESOFT program (Moilanen & Pedersen 2001), based on error estimates and corrections described in Pedersen & Boomsma (1999). b The uncorrected upper 95% confidence limit divided by (1 ? /); see also figure 1. c Calculated as l/we = (1 - Dest) X 1.0 + ?>est x ^pf, where ^^ = O.646 based on px = 0.77, p2 = 0.23 from Boomsma & Ratnieks (1996). d Calculated as r=0.25 + (0.5/mJ (Pamilo 1993). paternity skew (Pedersen & Boomsma 1999; Villesen et al. 1999). 3. RESULTS Not one doubly mated queen was detected in the 38 colonies analysed. Of special interest were the genotypes in the four 5. amabilis colonies that were previously ana- lysed by Murakami et al. (2000) and inferred to have had multiply mated queens. Our analysis allowed an unam- biguous rejection of this inference, as we could show that the two colonies that had elevated numbers of band differ- ences in the analysis by Murakami et al. (2000) each con- sisted of two genetically distinct groups of workers that originated from different queens, each mated to a single male. This was obtained by visual inspection of the geno- type distribution and was later confirmed by the MATE- SOFT program, which indicated that the genotypic patterns in these two colonies were in accordance with two unre- lated groups of offspring. These two polygynous colonies were therefore counted as four cases of single queen mat- ing in the re-analysis. Following the statistical procedures outlined above, we obtained maximum effective paternities (ms) ranging from 1.12 to 1.24 (table 1) and estimates of minimum intracolony relatedness of 0.65 to 0.69, corresponding to effective single mating in most if not all of the cases. The statistical analysis of the estimates of queen mating frequency is summarized in figure 1. The middle part of figure la shows the upper 95% confidence limits for the expected proportion of multiply mated queens in the five species analysed in the present study. The estimate varies somewhat due to differences in sample size, but shows that the upper confidence limit is usually less than 50% double mating. For comparison, we have plotted a series of similar analyses for three species of lower attine ants (the left side of figure la) and four species of leafcutter ants (the right side of figure la). These data were taken from published papers (Fjerdingstad et al. 1998; Fjerding- stad & Boomsma 2000; Bekkevold et al. 1999; Boomsma et al. 1999; Villesen et al. 1999), but had not yet been analysed for statistical confidence limits. When we calcu- lated the 95% confidence limits after pooling the data (assuming that each colony is a representative, random and independent sample of a distinct group of attine ants) into three overall categories (lower attines, higher non- leafcutting attines, leafcutter ants; figure lb), the differ- ences become even more convincing, because the larger sample sizes result in narrower confidence limits. 4. DISCUSSION Our analysis shows that five species of the higher attine genera Trachymyrmex and Sericomyrmex have singly mated queens. A much larger sample of colonies would be required to establish whether multiple mating is com- pletely absent or just rare, but the current analysis is suf- ficient to demonstrate that queen mating frequency is effectively very close to 1.0 in all species investigated. Although all species were collected in Panama, there is no reason to assume that they are not representative of their genera, as typical nesting habitats were variable across species and similar to the variation that occurs in South America (Weber 1966; Holldobler & Wilson 1990; T. R. Schultz, personal observation). The result for 5. amabilis is especially interesting because this species was inferred to have multiple queen mating by Murakami et al. (2000). Our current analysis of material from the same five nests showed that three colonies were headed by a lone, singly mated queen, whereas two colonies had two genetically Proc. R. Soc. Lond. B The transition between single and multiple mating in attine ants P. Villesen and others 02PB0145.5 (a) lower attines higher attines leafcutter ants 1.0 ih 08 c CO o c >>? c ? & 0.6 3 3 cr | * -a "O ?? u ^ 0 4 C3 i. Si n 3 a o II > & 0.0 <> o *?- O ^s ^ ^ ri <6 ^L% "%: "% "% ^\ (b) means per group 1.0 0.6 0.4 0.2 0.0 r o?o , %, ^ %. \ ^ %, % % % Figure 1. An overview of the best available estimates of queen mating frequency throughout the attine ants, giving the observed frequencies of multiple mating (dots) and their 95% CI (bars), corrected for combined errors ('/' in table 1). (a) The data per species, and (b) the pooled data per group. The central part of (a) includes both the data for the five species of non-leafcutting higher attines obtained in this study, and a series of similar data for three species of lower attine ants (at the left; from Villesen et al. 1999) and four species of Atta and Acromyrmex leafcutter ants (at the right; from Fjerdingstad et al. 1998; Bekkevold et al. 1999; Boomsma et al. 1999; Fjerdingstad & Boomsma 2000). (b) The same data pooled for the three major categories of fungus-growing ants (the lower attines, the higher non-leafcutting attines (this study) and the leafcutter ants. distinct groups of workers that originated from different queens, each mated to a single male. This indicates that facultative polygyny occurs in S. amabilis, perhaps simi- larly to the pattern observed in A. echinatior (Bekkevold et al. 1999), although occasional raiding of broods from a neighbouring colony could also explain the genotype dis- tributions observed. Further studies combining complete colony excavation and microsatellite genotyping will be needed to clarify this issue. Our results demonstrate that it is highly unlikely that the transition from single to multiple mating occurred before or coincident with the origin of the entire higher attine group (the four terminal genera in figure 2). To be supported, such a hypothesis would require the improb- able scenario that we have by chance sampled five species in two different genera, which have all secondarily lost multiple mating. It thus appears that multiple queen mat- ing originated in the common ancestor of Atta and Acro- myrmex (a group that may, due to the paraphyly of Trachymymiex discussed above, ultimately include one or a few additional Trachymymiex species, none of which were examined in this study). These novel insights con- cerning attine phylogeny (cf. figure 2) have made a third hypothetical transition (Villesen et al. 1999), intermediate between the two scenarios addressed here, obsolete. A relatively recent mating system transition in the higher attines implies that the origin of multiple mating in fungus-growing ants coincided with the specialization for leafcutting behaviour and not with the shift from non-spe- cialized fungal symbionts (utilized by all lower attines) to long term clonal symbionts (utilized by all higher attines) (see also final paragraph of ? 4). The occupation of the new herbivorous 'leafcutter' niche produced larger colon- ies with longer lived queens and more complex forms of social organization, and a higher 'visibility' to parasites, but the present data do not allow us to infer whether the mating system transition was causal or consequential in this rapid sequence of evolutionary developments. While it remains possible that future studies will reveal occasional double matings in some of the non-leafcutter attine gen- era, the phylogenetic correlation between the queen mat- ing system and particular social and ecological grades documented here (figure 2) indicates that it is unlikely that similar major transitions to multiply mated queens have occurred in other branches of the attine tree. The fact that multiple queen mating is restricted to the socially most derived clade of the attine ants has interest- ing parallels with phylogenetic patterns of multiple mating found in bees and wasps (Boomsma & Ratnieks 1996; Strassmann 2001). In the vespine wasps, multiple queen Proc. R. Soc. Lond. B 02PB0145.6 P. Villesen and others The transition between single and multiple mating in attine ants ] Q Myrmicocrypta O Apterostigma Mycocepurus ] Mycetophylax (part) 5 spieces ] Mycetarotes ] Mycetosoritis ] Q Cyphomyrmex Mycetophylax (part) 1 spieces O Sericomyrmex Trachymyrmex (part) 4 spieces 3 Q Trachymyrmex (part) 37 spieces Atta Figure 2. A simplified attine phylogeny based on Schultz (2000) and T. R. Schultz (unpublished data); see also Schultz & Meier (1995) and Wetterer et al. (1998). Numbers above branches are bootstrap values from maximum-likelihood analyses of DNA sequences from CO 1 and EF-1 a using the model GTR + / + gamma (see text). (White branches, lower attines; shaded branches, higher attines; black branches, leafcutter ants; single male sign, single mating; three male signs, multiple mating; arrow, possible origin of multiple mating.) The two mutually exclusive hypotheses for the transition from single to multiple queen mating tested in the present study are indicated by the arrows: (i) a hypothetical transition that occurred prior to, or coincident with, the origin of the higher attines; and (ii) a hypothetical transition that coincided with the origin of the leafcutter clade, which may include one or a few species currently still assigned to the paraphyletic genus Trachymyrmex. The results of the present study confirm the second hypothesis. See text for further details. mating is also restricted to the most derived genera, although there may have been two independent tran- sitions, one at the base of the genus Vespula and one within the Dolichovespula norwegica group (Foster et al. 1999). Interestingly, however, the latter (less derived) transition only includes partial multiple mating rather than the obligatory multiple mating present in the genus Ves- pula. Within the family Apidae, all species of honeybees (Apis) so far investigated have extremely high queen mat- ing frequencies (Estoup et al. 1994; Moritz et al. 1995; Palmer & Oldroyd 2000), but outside this derived clade, evolution towards multiple queen mating has been rare. One case of partial double mating has been documented in bumble-bees (Estoup et al. 1995; Schmid-Hempel & Schmid-Hempel 2000) and another isolated case exists in stingless bees (Paxton et al. 1999; Peters et al. 1999). It thus seems that evolutionary transitions from exclusively single mating to some combination of single-double or single-multiple mating have originated a number of times in bees and wasps, but that the species occupying these grades have not necessarily undergone further evolution- ary modification towards obligatory multiple mating with considerably higher queen mating frequencies. This over- all pattern of rarity of multiple mating was previously documented for ants by Boomsma & Ratnieks (1996). Since that review, a new case of multiple queen mating has been discovered in Pogonomyrmex harvester ants (Cole & Wiernasz 1999), confirming earlier observations by Holldobler (1976). Although there is some controversy over the interpretation of these results (Fjerdingstad & Keller 2000) because estimates of absolute and effective queen mating frequencies are lacking, it seems that the Pogonomyrmex case is similar to the leafcutter ants in that all queens mate with multiple males. Other recent studies on ants have shown exclusive single mating (Pheidole pallidula; Aron et al. 1999) or some combination of single and multiple mating: weaver ants Oecophylla smaragdina (Eraser & Crozier 1998) and three species of Formica ants (Walin et al. 1998). These studies all showed average effective queen mating frequencies close to unity (less than 1.04 for P. pallidula, 1.10 for O. smaragdina and 1.35- 1.47 for the Formica species). The data thus remain con- sistent with the hypothesis that multiple mating in ants is relatively rare, and that obligatory multiple mating is restricted to very few derived taxa, as inferred by Boomsma & Ratnieks (1996). Queens of monogynous large colonies could be sperm limited (Cole 1983), a factor sufficient to generate selec- tion for multiple mating. However, this mechanism seems unlikely to be very strong or general, because examples of species with large colonies and singly mated queens are common (Boomsma & Ratnieks 1996). In Atta colombica there is a significant increase in the amount of sperm stored by the queen and the number of males with which she mates (Fjerdingstad & Boomsma 1998). However, this effect might have arisen because, after the origin of Proc. R. Soc. Lond. B The transition between single and multiple mating in attine ants P. Villesen and others 02PB0145.7 multiple mating by queens, males started to mate multiply as well and were more discriminatory in allocating their sperm, forcing queens to obtain matings from an even greater number of males. Another argument against sperm limitation as a general explanation for the evolution of multiple queen mating is that queens of Acromyrmex octo- spinosus (Boomsma et al. 1999) and possibly Pogonomyr- mex occidentalis (Holldobler 1976; Cole & Wiernasz 1999) have high numbers of mates, but colony sizes at least an order of magnitude less than A. colombica (Wilson 1971; Wiernasz & Cole 1995). The leading hypothesis for the evolution of multiple queen mating has been that genetically diverse colonies suffer less from omnipresent parasites than genetically homogeneous colonies (Schmid-Hempel 1998). This hypothesis is supported by correlative comparative evi- dence for ants (Schmid-Hempel & Crozier 1999), and convincing specific evidence was recently obtained for bumble-bees (Baer & Schmid-Hempel 1999, 2001). The disease argument carries most force when addressing spe- cialized and well-adapted parasites. Of this category only one is known in the attine ants, a newly discovered fungal parasite of the genus Escovopsis, which attacks gardens of attine ants (Currie et al. 1999). This parasitic fungus is most prevalent in the higher attines, which may be due to a long evolutionary history of clonal propagation of the fungus (Chapela et al. 1994). The lower attines by con- trast have been shown to be able to acquire novel cultivar fungi from free-living sexual populations (Mueller et al. 1998) and lower attine colonies are perhaps therefore more robust against Escovopsis infection. If increased within-colony genetic variation in attine ants results in better defence against a specialized pathogen like Escov- opsis, targeted at an increasingly vulnerable symbiont, multiple queen mating might be selected for. This 'evol- utionary cost of long-term clonality' hypothesis is consist- ent with the fact that Escovopsis is present in all attine genera (Currie et al. 1999). However, all fungi cultivated by the higher attines (including Trachymyrmex and Serico- myrmex as well as the leafcutter ants) seem to share a simi- lar and equally long history of clonal propagation, as they all belong to the monophyletic Gl group of cultivars (Chapela et al. 1994). All other things being equal, there should thus be no reason for queen mating frequency to differ between Trachymyrmex and Sericomyrmex on the one hand, and Atta and Acromyrmex on the other, unless the previously mentioned differences in colony size, social complexity and longevity are also important. We con- clude, therefore, that we now have very good correlational evidence for a single major transition from single to mul- tiple queen mating in attine ants, but that additional research is required to identify the decisive causal factors that drove attine societies across this evolutionary bound- ary. We thank the Smithsonian Tropical Research Institute, Pan- ama, for providing logistic help and facilities to work in Gam- boa and the Institute Nacional de Recursos Naturales Renovables (INRENARE) for permission to sample ants in Panama. We thank Allen Herre, Jes S. Pedersen, David R. Nash and Boris Baer for comments on this manuscript. Jes S. Pedersen provided invaluable help with the statistical analysis and by making the MATESOFT program available. P.V. was sup- ported by a PhD grant from the Faculty of Natural Sciences of the University of Aarhus. T.M. was supported by a grant from the Japan Society for the Promotion of Science Research Fellowships for Young Scientists. T.R.S. was supported by National Science Foundation grants DEB 9707209 and DEB 0110073, and by Smithsonian Scholarly Studies grant no. 140202-3340-410. J.J.B. was supported by grants from the Danish Natural Science Research Council and the Carlsberg Foundation. REFERENCES Adams, R. M. M., Mueller, U. G., Holloway, A. K., Green, A. M. & Narozniak, J. 2000 Garden sharing and garden stealing in fungus-growing ants. Naturwissenschaften 87, 491-493. Aron, S., Campan, E., Boomsma, J. & Passera, L. 1999 Social structure and split sex ratios in the ant Pheidole pallidula. Ethol. Ecol. Evol. 11, 209-227. Baer, B. & Schmid-Hempel, P. 1999 Experimental variation in polyandry affects parasite loads and fitness in a bumble- bee. Nature 397, 151-154. Baer, B. & Schmid-Hempel, P. 2001 Unexpected conse- quences of polyandry for parasitism and fitness in the bum- ble-bee B. terrestris. Evolution 55, 1639-1643. Bekkevold, D., Frydenberg, J. & Boomsma, J. J. 1999 Multiple mating and facultative polygyny in the Panamanian leaf- cutter ant Acromyrmex echinatior. Behav. Ecol. Sociobiol. 46, 103-109. Beshers, S. & Traniello, J. A. 1994 The adaptiveness of worker demography in the attine ant Trachymyrmex septentrionalis. Ecology 75, 763-775. Boomsma, J. J. & Ratnieks, F. L. W. 1996 Paternity in eusocial Hymenoptera. Phil. Trans. R. Soc. Lond. B 351, 947-975. Boomsma, J. J., Fjerdingstad, E. J. & Frydenberg, J. 1999 Mul- tiple paternity, relatedness and genetic diversity in Acromyr- mex leafcutter ants. Proc. R. Soc. Lond. B 266, 249-254. (DOI 10.1098/rspb.l999.0629.) Chapela, I., Rehner, S., Schultz, T. R. R. & Mueller, U. G. 1994 Evolutionary history of the symbiosis between fungus- growing ants and their fungi. Science 266, 1691-1694. Cole, B. J. 1983 Multiple mating and the evolution of social behavior in the hymenoptera. Behav. Ecol. Sociobiol. 12, 191-201. Cole, B. J. & Wiernasz, D. C. 1999 The selective advantage of low relatedness: growth in the harvester ant, Pogonomyrmex occidentalis. Science 285, 491-493. Currie, C, Mueller, U G. & Malloch, D. 1999 The agricul- tural pathology of ant fungus gardens. Proc. Natl Acad. Sci. USA 96, 7998-8002. Estoup, A., Solignac, M. & Cornuet, J. M. 1994 Precise assess- ment of the number of patrilines and of genetic relatedness in honeybee colonies. Proc. R. Soc. Lond. B 258, 1-7. Estoup, A., Scholl, A., Pouvreau, A. & Solignac, M. 1995 Monoandry and polyandry in bumble-bees (Hymenoptera; Bombinae) as evidenced by highly variable microsatellites. Mol. Ecol. 4, 89-93. Felsenstein, J. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39, 783-791. Fjerdingstad, E. J. & Boomsma, J. J. 1998 Multiple mating increases the sperm stores of Atta colombica leafcutter ant queens. Behav. Ecol. Sociobiol. 42, 257-261. Fjerdingstad, E. J. & Boomsma, J. J. 2000 Queen mating fre- quency and relatedness in young Atta sexdens colonies. Insectes Sociaux 47, 354-356. Fjerdingstad, E. J. & Keller, L. 2000 The nature of ant colony success. Nature 287, 1363b. Fjerdingstad, E. J., Boomsma, J. J. & Thoren, P. 1998 Multiple paternity in the leafcutter ant Atta colombica: a microsatellite DNA study. Heredity 80, 118-126. Proc. R. Soc. Lond. B 02PB0145.8 P. Villesen and others The transition between single and multiple mating in attine ants Foster, K., Seppa, P., Ratnieks, F. & Thoren, P. 1999 Low paternity in the hornet Vespa crabro indicates that multiple mating by queens is derived in vespine wasps. Behav. Ecol. Sociobiol. 46, 252-257. Fowler, G. H., Pereira-da-Silva, V., Forti, L. C. & Saes, N. B. 1986 Population dynamics of leaf-cutting ants: a brief review. In Fire ants and leaf cutting ants: biology and manage- ment (ed. C. S. Lofgren & R. K. Vander Meer), pp. 123- 145. Boulder, CO: Westview Press. Fraser, V. S. & Crozier, R. H. 1998 The mating structure of Oecophylla smaragdina ants. In Social insects at the turn of the millennium. Proc. XIII International Congress oflUSSI (ed. M. P. Schwarz & K. Hogendoorn), p. 172. Adelaide, Australia: Flinders University Press. Hamilton, W. D. 1964 The genetical evolution of social behaviour, vol. 2.J. Theoret. Biol. 7, 17-52. Hillis, D. M. & Bull, J. J. 1993 An empirical test of bootstrap- ping as a method for assessing confidence in phylogenetic analysis. Syst. Biol. 42, 182-192. Holldobler, B. 1976 The behavioral ecology of mating in har- vester ants (Hymenoptera: Formicidae: Pogonomyrmex). Behav. Ecol. Sociobiol. 1, 405-423. Holldobler, & Wilson, E. O. 1990 The ants. Berlin: Springer. Moilanen, A. & Pedersen, J. S. 2001 MATESOFT: a program for genetic analysis of mating systems 0.99b. Division of Population Biology, University of Helsinki, Finland. Moritz, R. A., Kryger, P., Koeniger, G., Koeniger, N., Estoup, A. & Tingek, S. 1995 High degree of polyandry in Apis dor- sata queens detected by DNA microsatellite variability. Behav. Ecol. Sociobiol. 37, 357-363. Mueller, U. G, Schultz, T. R., Currie, C. R., Adams, R. M. M. & Malloch, D. 2001 The origin of the attine ant- fungus mutualism. Q. Rev. Biol. 76, 169-197. Mueller, U. G, Rehner, S. & Schultz, T. R. 1998 The evol- ution of agriculture in ants. Science 281, 2034-2038. Murakami, T. 1998 Sociobiological studies of fungus-growing ants Attini: effects of insemination frequency on the social evolution. PhD thesis, Hokkaido University, Japan. Murakami, T., Higashi, S. & Windsor, D. 2000 Mating fre- quency, colony size, polyethism and sex ratio in fungus- growing ants (Attini). Behav. Ecol. Sociobiol. 48, 276-284. Ortius-Lechner, D., Gertsch, P. J. & Boomsma, J. J. 2000 Vari- able microsatellite loci for the leafcutter ant Acromyrmex echi- natior and their applicability to related species. Mol. Ecol. 9, 114-116. Palmer, K. A. & Oldroyd, B. P. 2000 Evolution of multiple mating in the genus Apis. Apidologie 31, 235-248. Pamilo, P. 1982 Multiple mating in Formica ants. Hereditas 97, 37-45. Pamilo, P. 1993 Polyandry and allele frequency differences between the sexes in the ant Formica aquilonia. Heredity 70, 472-480. Paxton, R. J., Weissschuh, N., Engels, W. & Hartfelder, K. 1999 Not only single mating in stingless bees. Naturwissen- schaften 86, 143-146. Pedersen, J. S. & Boomsma, J. J. 1999 Multiple paternity in social Hymenoptera: estimating the effective mate number in single-double mating populations. Mol. Ecol. 8, 577-587. Peters, J., Queller, D., Imperatriz-Fonseca, V., Roubik, D. & Strassmann, J. 1999 Mate number, kin selection and social conflicts in stingless bees and honeybees. Proc. R. Soc. Lond. B266, 379-384. (DOI 10.1098/rspb.1999.0648.) Rodriguez, F., Oliver, J. L., Marin, A. & Madina, J. L. 1990 The general stochastic model of nucleotide substitution. J. Theor. Biol. 142, 485-501. Schmid-Hempel, P. 1998 Parasites in social insects. Monographs in behavior and ecology. Princeton University Press. Schmid-Hempel, P. & Crozier, R. H. 1999 Polyandry versus polygyny versus parasites. Phil. Trans. R. Soc. Lond. B 354, 507-515. (DOI 10.1098/rstb.l999.0401.) Schmid-Hempel, R. & Schmid-Hempel, P. 2000 Female mat- ing frequencies in Bombus spp. from Central Europe. Insectes Sociaux 47, 36-41. Schultz, T. R. 2000 The origin and evolution of fungus-grow- ing: clues from the phylogeny of the fungus-growing ants. In Abstracts from the International Congress of Entomology (21: 2000: Foz do Iguassu), vol. 2, p. 865. Londrina, Brazil: Embrapa Soja. Schultz, T. R. & Meier, R. 1995 A phylogenetic analysis of the fungus-growing ants (Hymenoptera: Formicidae: Attini) based on morphological characters of the larvae. Syst. Ento- mol. 20, 337-370. Strassmann, J. E. 2001 The rarity of multiple mating in the social Hymenoptera. Insectes Sociaux 48, 1-13. Swofford, D. L. 2001 . Paup*. Phylogenetic analysis using parsi- mony ("and other methods), v. 4. Sunderland, MA: Sinauer. Urich, F. W. 1895 Notes on the fungus growing and eating habit of Sericomyrmex opacus, Mayr. Trans. Entomol. Soc. Lond. 77-78. Villesen, P., Gertsch, P., Frydenberg, J., Mueller, U. G & Boomsma, J. 1999 Evolutionary transition from single to multiple mating in fungus-growing ants. Mol. Ecol. 8, 1819-1825. Villesen, P., Gertsch, P. J. & Boomsma, J. J. 2002 Microsatel- lite primers for fungus-growing ants. Mol. Ecol. Notes. (In the press.) Walin, L., Sundstrom, L., Seppa, P. & Rosengren, R. 1998 Worker reproduction in ants: a genetic analysis. Heredity 81, 604-612. Walsh, P. S., Metzger, D. A. & Higuchi, R. 1991 Chelex 100 as a medium for simple extraction of DNA for PCR-based typing from forensic material. BioTechniques 10, 506-513. Weber, N. A. 1945 The biology of the fungus-growing ants. Part VIII. The Trinidad, B. W. I., species. Revista de Entom- ologia Rio de Janeiro 16, 1-88. Weber, N. A. 1966 Fungus-growing ants. Science Wash. DC 153, 587-604. Weber, N. A. 1972 Gardening ants: the Attines. Mem. Am. Phil. Soc. 92, 1-138. Weber, N. A. 1979 Fungus-culturing by ants. In Insect-fungus symbiosis: mutualism and commensalism (ed. L. R. Batra), pp. 77-116. Montclair, NJ: Allanheld, Osmun. Wetterer, J., Schultz, T. R. & Meier, R. 1998 Phylogeny of fungus-growing ants (tribe attini) based on mtDNA sequence and morphology. Mol. Phylogenet. Evol. 9, 42-47. Wiernasz, D. C. & Cole, B. J. 1995 Spatial distribution of Pogonomyrmex occidentalis: recruitment, mortality and over- dispersion. J. Anim. Ecol. 64, 519-527. Wilson, E. O. 1971 The insect societies. Cambridge, MA: Belknap. Zar, J. H. 1999 Biostatistical analysis, 4th edn. Upper Saddle River, NJ: Prentice-Hall. As this paper exceeds the maximum length normally permitted, the authors have agreed to contribute to production costs. Proc. R. Soc. Lond. B