Abstract In many polygynous ant species, established colonies adopt new queens secondarily. Conflicts over queen adoption might arise between queens and.
© Springer-Verlag 1996
Behav Ecol Sociobiol (1996) 39 : 275–284
Jay D. Evans
Queen longevity, queen adoption, and posthumous indirect fitness in the facultatively polygynous ant Myrmica tahoensis
Received: 20 February 1996 /Accepted after revision: 25 May 1996
Abstract In many polygynous ant species, established colonies adopt new queens secondarily. Conflicts over queen adoption might arise between queens and workers of established colonies and the newly mated females seeking adoption into nests. Colony members are predicted to base adoption decisions on their relatednesses to other participants, on competition between queens for colony resources, and on the effects that adopted queens have on colony survivorship and productivity. To provide a better understanding of queen-adoption dynamics in a facultatively polygynous ant, colonies of Myrmica tahoensis were observed in the field for 4 consecutive years and analyzed genetically using highly polymorphic microsatellite DNA markers. The extreme rarity of newly founded colonies suggests that most newly mated queens that succeed do so by entering established nests. Queens are closely related on average (0 = 0.58), although a sizable minority of queen pairs (29 %) are not close relatives. An experiment involving transfers of queens among nests showed that queens are often accepted by workers to which they are completely unrelated. Average queen numbers estimated from nest excavations (harmonic mean = 1.4) are broadly similar to effective queen numbers inferred from the genetic relatedness of colony members, suggesting that reproductive skew is low in this species. Queens appear to have reproductive lifespans of only 1 or 2 years. As a result, queens transmit a substantial fraction of their genes posthumously (through the reproduction of related nestmates), in comparison to direct and indirect reproduction while they are alive. Thus queens and other colony members should often J.D. Evans Department of Biology, University of Utah, Salt Lake City, UT 84112, USA J.D. Evans (*) Department of Entomology, University of Georgia, Athens, GA 30602, USA
accept new queens when doing so will increase colony survivorship, in some cases even when the adopted queens are not close relatives. Key words Polygyny · Eusociality · Life history · Microsatellite DNA · Formicidae
Introduction Social insect colonies often contain two or more fertile queens (polygyny : Hölldobler and Wilson 1990; Keller 1993a). Polygyny is associated with a host of physiological (Keller and Passera 1989), ecological (Herbers 1993), and life-historical (Elmes 1980) patterns. In social insects generally, polygynous queens are smaller than their monogynous counterparts. In polygynous ants, these differences in size, and corresponding differences in energy reserves, are correlated with a tendency among queens to seek adoption into existing nests (secondary colony founding), rather than attempt to initiate nests without the benefits of an established worker force (Keller and Passera 1989). Queen adoption by ant colonies will affect the fitness of entering queens, established queens, and the workers of established nests. Approaching queens are almost certain to increase their fitness by entering established nests, given the apparently low levels of de novo colony establishment by queens. The impact of an adopted queen on the fitnesses of her new nestmates should depend on the interplay between their relatedness to the adopted queen and her effects on colony productivity (Nonacs 1988; Pamilo 1991), as well as on the degree to which she competes with existing colony members for reproduction (e.g., Keller and Reeve 1994). In this paper, I use a kin-selection approach toward understanding the implications of queen adoption for ant colony members, then present data on the frequency and impact of queen adoption in colonies
276
of the subalpine ant Myrmica tahoensis. This approach will help to establish whether adopted queens increase the inclusive fitness of those adopting them or, alternatively, whether they act as parasites on established colonies (Elmes 1973). Theorists have long understood that the strategic maneuvers of workers and queens may have long-term as well as immediate effects on the inclusive fitnesses of the participants (Franks et al. 1990; Nonacs 1988, 1993). An important implication is that instantaneous measures of relatedness and reproductive success may not be sufficient to evaluate the full effects of a colony member’s behavior (Keller 1993b). In the context of intracolonial conflicts, Franks et al. (1990) proposed that queens in colonies of the ant Harpagoxenus sublaevis might tolerate fertile workers, when doing so allows these workers to produce a large flush of males in the year after the queen dies. Similarly, an increase in the “posthumous” fitness of current colony members could explain the tendency of ant colonies to adopt queens secondarily, despite the fact that adopted queens are likely to compete with existing colony members. Few empirical studies have directly addressed the issue of long-term reproductive strategies in social insects, mainly because of the difficulties involved in attempting to measure the relevant variables. Here I describe a detailed multi-year study of colony structure in the facultatively polygynous ant Myrmica tahoensis. In many respects, colonies of M. tahoensis are ideal for such a study. Colonies vary considerably in queen number (from 0 to 8) and in levels of relatedness among nestmates. Workers are fertile (Evans 1993) and may therefore compete with each other and with their queens for the production of males. Colonies will accept alien queens and larvae (Evans 1995), so it is possible to manipulate relatedness levels and queen numbers experimentally. Finally, highly polymorphic microsatellite DNA markers developed for this species allow the relatednesses of nestmates to be estimated with high precision (Evans 1993, 1995). Genetic analyses at these loci therefore provide an unusually clear look at the relatedness patterns and reproductive dynamics of colonies. Since these analyses can be conducted with minute amounts of DNA, the genotypes of all colony members, not just adults, can be determined. After summarizing demographic data for colonies, I describe important parameters of the breeding system (sensu Ross 1993) of M. tahoensis. Estimates of genetic relatedness between nestmate queens, brood, and workers are used to estimate the effective number of queens within colonies. Rates of queen turnover within mature colonies of M. tahoensis are estimated from differences in relatedness among colony members of different ages. When queens die, they may be replaced either by daughter queens or by unrelated queens produced by other nests. Individual queen pairs are
compared to determine both the average relatedness among queens and the variance in queen-queen relatedness. The variance in queen-queen relatedness gives an estimate of the frequency at which unrelated queens are adopted by established nests. The readiness of established nests to accept outside queens is also assessed by transferring queens among nests in the laboratory. Patterns of relatedness, reproductive skew, and queen turnover, in conjunction with the long development times observed in M. tahoensis, suggest that workers and queens in this species realize a substantial proportion of their direct and indirect reproduction posthumously. Adopted queens might help to ensure the realization of this posthumous reproduction, perhaps compensating for the fact that adopted queens compete with existing colony members for reproduction (Nonacs 1988). When the effects of adopted queens on colony survivorship are large, as is likely for M. tahoensis, colonies might benefit from queen adoption even when adopted queens are not close relatives.
Methods Species and study population Myrmica tahoensis occurs in montane and subalpine habitats throughout much of western North America (Smith 1979). As with most species of Myrmica (Elmes 1980), colonies are facultatively polygynous (Evans 1995). In Colorado, M. tahoensis occurs commonly in small subalpine meadows surrounded by aspen and sprucefir forest. Colonies live almost exclusively under stones, usually within 0.5 m of the soil surface. For this study, field colonies of M. tahoensis were collected from seven meadows near the Rocky Mountain Biological Laboratory (RMBL), Gunnison County, Colorado (2900 m elevation). Voucher specimens have been deposited at the Museum of Comparative Zoology (Harvard University) and in the collections of A. Francoeur (University of Québec, Chicoutimi) and the author.
Collection and demography of nests Complete nests (n = 189) were excavated immediately after sexual males and females had enclosed in late July and early August 1991–1993. Nests were dug near midday, when most colony members were near the surface. This fact, and the relatively small size of colonies (generally 100 – 400 workers), allowed for a fairly complete census of all nestmates. Colonies were placed into plastic bags, then were sorted on large plastic tarpaulins divided into squares separated by barriers of Tree Tanglefoot (Tanglefoot Co., Grand Rapids, Mi.). All colony members were counted, then stored individually in 95 % ethanol in microcentrifuge tubes. An additional group of 53 nests was collected in mid-June 1992–1994. These nests were established in the laboratory for behavioral trials (Evans 1996), after which individuals were saved for genetic analyses. At two sites, Romar and Bear, all colonies and available nest sites (stones) were mapped and tagged in 1991. These sites were then re-surveyed in 1992 and 1993, to measure colony disappearance rates and the colonization of uninhabited nest sites. Workers and brood were collected from a subsample of these colonies each year, to provide estimates of relatedness and colony-level genetic turnover between years.
277 DNA amplification and genetic analyses Genotypes at microsatellite loci Myrt2, Myrt3, and Myrt4 (Evans 1993) were determined from DNA extracted from 1177 individuals in 101 colonies. These loci contain tandem repeats of the nucleotide pair CA /GT. Alleles at each locus differ from each other in the number of CA/GT repeats they contain, as measured by size-separation on electrophoretic gels. DNA was first extracted from adults and pupae as described (Evans 1995). Individual eggs (post-hatching embryos) and larvae were homogenized in 50 or 500 µl TE[3 (1: 1000 dilution of TE; Sambrook et al. 1989), respectively. The homogenate was diluted 1:10 for genotypic analyses. Individual loci were then amplified by the polymerase chain reaction (PCR) and fragments were analyzed by acrylamide-gel electrophoresis. Allele sizes were determined by comparison to known size standards (Evans 1995). Loci Myrt2, Myrt3, and Myrt4 showed 24, 26 and 40 alleles, respectively, in M. tahoensis. Rare alleles of similar size were combined, giving a maximum of 25 recognized alleles per locus. Regression estimates based on shared alleles (Queller and Goodnight 1989) were then used to characterize the average relatedness among colony members. These estimates (along with corresponding standard error estimates produced by jackknifing over colonies) were made using the computer program Relatedness 4.2, obtained from D. Queller (Rice University). Deme-specific allele frequencies, used in the relatedness calculations, were generated for each of the seven meadows. To allow comparisons of their relatedness to female pupae and workers, only female eggs and larvae (as classified by heterozygosity at one or more polymorphic loci) were used in these analyses. Given the three highly polymorphic loci (observed heterozygosities of 0.69, 0.72, and 0.86 for the three screened loci in adult females), nearly all females are expected to be heterozygous at one or more locus. A total of 116 female larvae from 25 colonies were included in the analysis, along with 50 eggs from ten colonies.
Two levels of male relatedness were used to bracket the range of plausible values of rm (see Seppä 1994). At one extreme, rm = 0, the mates of all queens in the colony are assumed to be unrelated. Alternatively, the mates of co-occurring queens might themselves be nestmates, in which case rm should reflect the observed level of relatedness among male nestmates (0 = 0.37, SE = 0.04, n = 120 males in 23 colonies).
Queen turnover Differences in relatedness among colony members of different ages can be used to generate an estimate of year-to-year genetic turnover. In colonies collected during late summer, worker pupae are likely to be 1 year younger than adult workers, since workers appear to be produced in one major flush in this species (J.D. Evans, unpublished work). Similarly, worker pupae are likely to be a full year older than eggs and small larvae, since M. tahoensis brood generally take two seasons to develop (Fig. 1; Brian and Jones 1980, for M. rubra). Relatedness of adult workers to eggs and larvae provides a basis for estimating genetic turnover over a 2-year interval. In order to estimate genetic turnover in colonies, genetic similarity between two age groups (S) was defined as the observed relatedness between members of the two groups divided by their expected relatedness on the basis of relatedness levels within each group : S = 2rxy/(rxx + ryy)
where x and y are members of two age classes. If x and y belong to a homogeneous group, S should approach unity. This calculation is not biased by differences in relatedness found within each age class. Changes in relatedness were then used to predict the yearly turnover of queens in colonies. The relationship between genetic turnover, queen survivorship, and the relatedness of adopted queens to current colony members can be characterized as follows : S = (P) + (1[P)raq
Breeding system Population-genetic structure, in the form of Fis and Fst (Wright 1951), was estimated from allele and genotype distributions within and across the seven sites. Standard errors for these estimates were generated by jackknifing across colonies or sites, using Relatedness 4.2. Inbreeding in M. tahoensis was calculated using two methods. First, Fis was estimated for each locus, using allele and genotype frequencies from each deme (meadow). Some microsatellite loci include null alleles (Pemberton et al. 1995) that could inflate inbreeding estimates. To avoid this potential bias, a pedigree-based estimate of inbreeding was also generated. Full-sib genotypes were used to infer the genotypes of each queen and her mate. Given these likely parental genotypes (from 26 colonies with seven or more genotyped full sibs), relatedness estimates were calculated for each queen and her mate using the algorithms in Queller and Goodnight (1989). These estimates then served as a second predictor of Fis. Effective queen numbers (Ne) were estimated from larval, pupal, and worker relatedness estimates by means of the equation : Ne = (rq + 2rm[4rs)/(rq + 2rm[4rf)
(4)
where P = the average proportion of queens surviving year-to-year, raq = the relatedness of adopted queens to existing queens, and S = the average genetic similarity between colony members in two consecutive years. When all queens survive from one year to the next, S = 1 and there is no decrease in genetic relatedness between colony members. When queens die, or when the relative contributions of queens change from one year to the next, S will decrease, reaching zero in the extreme circumstance that queens reproduce only one year after which they are replaced by unrelated queens.
(1)
where rq is the relatedness among nestmate queens, rm is the average relatedness of the mates of nestmate queens, rs is the relatedness among full sisters and rf is the relatedness estimate for all nestmate females (Ross 1993). Pedigree analyses in single-queen colonies strongly suggest that queens mate only once (Evans 1993 and unpublished work). Assuming single mating, and using inferred inbreeding levels (described in the results), sisters are predicted to be related by 0 = approximately 0.80. This implies that Ne = (rq + 2rm[3.2)/(rq + 2rm[4rf)
(3)
(2)
Fig. 1 Age structure of Myrmica tahoensis colonies collected in late July (t = 0) (Sp0 spring of same year, Sp–1 spring of previous year, Sp–2 spring 2 years prior, dashed line refers to larval period, solid line adult period). Data based on field collections for M. tahoensis and records collected for European species of Myrmica (Brian 1983)
278 When queens reproduce at equal rates, Eq. 4 can be solved for the proportion of queens that survive from one year to the next : P = (S[raq)/ (1[raq)
(5)
Immediate and future inclusive fitness gained by queens A simple kin-selection model was developed to explore the relative importance of immediate and future reproduction to a colony’s existing queens. The expected direct fitness received by a queen in 5 years was defined as: 4
] Pi/N
(6)
i=0
where i = 0 is the current year, Pi is her likelihood of being alive in a particular year, and N is the (effective) number of queens in a nest. The indirect fitness received by a queen was defined by :
queen status (queenless, monogynous, or polygynous) of receiving nests. The factors in the analysis were colony type, queen (random factor), and host nest (random factor nested in colony type). For each trial, results were scored as 0 (colony rejected queen) or 1 (colony accepted queen). Accepted queens (1) remained in the brood area of their host colonies and (2) were not treated aggressively by workers after a period of several hours. Behavioral trials conducted on these colonies suggest that accepted queens show similar survivorship and fecundity when compared to other polygynous queens (Evans 1996). These experiments were conducted using established queens, rather than newly mated females (gynes) that are arguably most likely to seek entry into established nests. This creates a potential bias, but there is no a priori reason to expect a different response by host colonies to newly mated or older non-relative queens. In fact, Elmes (1980) noted that solitary queens can be found moving outside of nests year-round in populations of M. rubra, suggesting that such queens are seeking entry into established nests.
4
] [(N[Pi)/N][rqPi+S i(1[p)i]
(7)
i=0
where rq is queen-queen relatedness and S, as before, reflects the genetic similarity of colony members from one year to the next. The value (N-pi) describes the proportion of a colony’s reproduction not produced by the focal queen, for year i. The function rq pi accounts for indirect fitness gained through nestmate queens still present in year i, while S i(1[P)i accounts for adopted queens (whose relatedness to the focal queen is predicted to be S i). Queen numbers and colony productivity are expected to remain the same, on average, through time. These estimates were then used in two different ways. First, the direct and indirect fitness returns gained by a queen in her last year of reproduction were compared to her expected indirect fitness gains from the four subsequent years. Second, a queen’s average total fitness gained while alive (directly and indirectly) was compared to that expected in the years after her death. This second analysis was limited to a total of five years, since queens in M. tahoensis are unlikely to impact the fitness of their colonies (either directly or indirectly) for a period longer than five years. Queen survivorship was allowed to vary from 0.0 to 0.6 per year, since the genetic turnover data suggest a maximum survivorship rate of 0.6 for queens. In both analyses, queen fitness returns were compared for colonies with average queen numbers of 1.5, 3, and 10. Genetic similarity, S, was fixed at 0.6, a value generated empirically for M. tahoensis.
Results
Queen introductions
Queen-queen relatedness
Genetic analyses suggested that unrelated queens are often adopted into established M. tahoensis nests. To assess the ability of established nests to discriminate unrelated queens from nestmate queens, non-nestmate queens were experimentally introduced into laboratory colonies. In July 1993, I conducted 104 such introductions with 24 queens and 25 recipient nests (some queens and nests were tested repeatedly). Queens were introduced into host colonies that originated at least 10 m distant from their own colonies. In addition, the autonomy of host and donor nests was established through behavioural experiments involving workers; “donor” colony workers were always attacked by workers from the recipient colony. After introductions, queens were re-introduced to their native nests, to determine whether their absence, or the experimental trial itself, led workers to reject them. Individual queens were introduced into from one to seven nests during the experiment, allowing an analysis of the relative abilities of individual queens to successfully enter established nests. Further, several colonies were used multiple times as hosts. Accordingly, a repeated-measures analysis of variance (Potvin et al. 1990) was used to distinguish between the effects of introduced queens, the nests into which they were placed, and the
Queens were related, on average, by 0 = 0.58 (SE = 0.06, n = 81 queens in 32 colonies). Queen-queen relatedness estimates were highly variable, ranging from 0 to 1.0 (Fig. 2). Some of this variance may result from sampling error. For example, queen pairs sharing no alleles at the three loci used might show low levels of relatedness if genotyped at a larger number of loci. Nevertheless, 22 queen-pairs (29 %) showed microsatellite alleles that positively excluded their being either full sisters or mother-daughter pairs. This result suggests that unrelated, or distantly related, queens are regularly adopted by nests. Queen-queen relatedness remained constant in relation to queen number (Kruskal-Wallis test, C 2 = 0.77, df = 4, ns; 0 = 0.57, 0.66, and 0.56 for colonies with two, three, and five queens, respectively).
Colony sizes Colonies collected immediately prior to their mating flights contained, on average, 1.12 queens (SE = 0.01) and 175 workers (SE = 18.4). Of 131 colonies 49 (37%) were apparently queenless when collected. Of 82 queenright colonies 39 (48 %) contained one queen, 30 contained two queens and 7, 3, and 3 colonies contained three, four and five queens, respectively (x = 1.79, SE = 0.11). Queen number and worker number were positively correlated (Spearman ρ = 0.38, n = 79, p < 0.00001). Genetic data (below) suggest that queenless nests have effective queen numbers similar to those found in the rest of the population. The harmonic mean queen number for all queenright nests (x = 1.42, SE = 0.06) therefore provides an appropriate predictor of colony-level relatedness across the population (Queller 1993).
279
Fig. 2 Estimated relatedness between all pairs of nestmate queens, and colony-level estimates of queen-queen relatedness
Fig. 3 Relatedness among workers in colonies of M. tahoensis. Data represent those colonies for which five or more workers were genotyped (n = 43)
Queen-mate relatedness F-statistics calculated for brood, queens, and workers of M. tahoensis suggest that inbreeding occurs at a low but significant level (Table 1). For all females in colonies, Fis = 0.18 ( jackknife SE = 0.03). Similarly, the estimated relatedness between females and their mates predicts that Fis = 0.16 (SE = 0.05). At this level of inbreeding, full sisters are expected to be related by 0 = 0.79–0.81, rather than r = 0.75 (Hamilton 1972; Pamilo 1985). Relatedness of female colony members Workers were related to each other by 0 = 0.56 (n = 447 individuals in 58 colonies, SE = 0.03). In 43 colonies for which five or more workers were genotyped, relatedness estimates ranged from 0.20 to 0.90 (Fig. 3). Worker pupae were related by 0 = 0.59 (SE = 0.06, n = 106 individuals in 22 colonies). Relatedness among 116 female larvae from 25 nests (0 = 0.45, SE = 0.04) tended to be lower than relatedness among workers, although this result was not statistically significant (two-tailed t-test, t = 1.7, df = 81, 0.10 > P > 0.05). Both larval and worker genotypes were estimated in 21 nests. In these nests, larval relatedness again Table 1 Estimates of Fis and Fst for three microsatellite loci. Allele and genotype frequencies were generated for a total of 744 queens, worker pupae, workers, gynes and males from seven sites (demes). Standard errors were estimated by jackknifing across colonies for Fis and across demes for Fst
tended to be lower than worker relatedness (0 = 0.42, SE = 0.03, versus 0 = 0.52, SE = 0.05). Relatedness among all larvae, including those that showed only one allele at all screened loci, was 0.42 (SE = 0.05, n = 154 individuals). Effective queen numbers were calculated from the relatedness of female offspring, the degree of inbreeding, and queen-queen relatedness (Table 2). Estimates of effective queen number were broadly similar, on average, to the physical queen numbers (harmonic mean) found at the time of collection for colonies. However, worker-worker relatedness did not correlate strongly with physical queen number at the individual colony level (Kruskal-Wallis test for queenless, monogynous and polygynous nests, X 2 = 2.7, df = 2, ns). Colony and queen turnover No solitary egg-laying queens were observed in 4 years of survey. This finding suggests that new nests are rarely founded independently by newly mated gynes. Instead, colonies appear to persist by continually adopting new queens and to reproduce, in large part, by budding. The rate at which existing queens are replaced can be inferred from differences in relatedness among colony
Locus
Heterozygosity expected (observed)
Fis (SE)
Fst (SE)
Myrt2 Myrt3 Myrt4
0.94 (0.69) 0.92 (0.72) 0.95 (0.86)
0.28 (0.05) 0.21 (0.04) 0.12 (0.03)
0.007 (0.01) 0.007 (0.01) 0.006 (0.01)
Total
–
0.18 (0.03)
0.007 (0.01)
280 Table 2 Effective queen numbers in Myrmica tahoensis colonies, as estimated from larval, pupal and adult worker genotypes. The parameter rm= relatedness between the mates of nestmate queens Group
No. individuals (no. colonies)
Relatedness x ± (SE)
Effective queen no., rm= 0
Effective queen no., rm= 0.37
Larvae Worker pupae Adult workers
116 (25) 106 (22) 447 (58)
0.45 (0.06) 0.59 (0.05) 0.56 (0.03)
2.2 1.5 1.6
3.9 1.7 1.9
Table 3 Genetic relatedness for colony members of different age classes (E eggs, L larvae, WP worker pupae, W workers). Individuals of different ages were sampled from colonies concurrently. Relatedness was estimated within each age class (i.e. raa and rbb, with the subscripts representing age classes 1 and 2), then was averaged across the two age classes (SEs shown in parentheses). Pair
E-W L-W L-WP WP-W Average
Individuals (colonies)
126 279 124 273 –
(11) (25) (16) (24)
This average was compared to the estimated relatedness between members chosen from each of the age classes (rab). The ratio between these values (an estimate of genetic turnover) is also presented, and the differences between the estimates is compared with a one-way t-test, P-values in table
Relatedness within age class 0.5 (raa+rbb)
Relatedness between age classes (rab)
(P-value)
0.39 0.51 0.52 0.57
0.08 0.20 0.38 0.42
0.21 0.39 0.73 0.73
–
(0.06) (0.06) (0.04) (0.04)
(0.02) (0.04) (0.04) (0.05)
–
members of different ages. Eggs, larvae, worker pupae, and workers collected from colonies all showed higher levels of within-age-group than between-group relatedness, indicating substantial genetic turnover between these age classes (Table 3). After compensating for age differences (1 or 2 years, Fig. 1), between-class relatedness comparisons indicate that colony members are approximately 60 % as closely related to nestmates in the next year’s brood as they are to members of their own age cohort (Table 3). Genetic similarity in a given nest across years reflects the pattern of queen mortality and the relatedness of existing queens to newly adopted queens (Fig. 4). The high rate of genetic turnover within colonies implies that M. tahoensis queens have very short lifespans. Even when all newly recruited queens are nonrelatives, only 60 % of all queens are predicted to survive from one year to the next. Two plausible estimates of relatedness between existing and adopted M. tahoensis queens are presented in Fig. 4. Gynes produced by colonies are related to existing queens by 0 = 0.4 (SE = 0.12, n = 9 colonies). This estimate provides an upper bound for relatedness of resident queens to adopted queens, which would be achieved if all accepted queens were daughters. If (as indicated by the queen-queen relatedness distribution. Fig. 1) unrelated or distantly related queens are occasionally adopted, the average relatedness between gynes and existing queens would be lower, perhaps as low as r = 0.2 (when half of the adopted queens are non-relatives). Using these two estimates as extremes, annual queen survivorship is predicted to be between 0.35 and 0.5
0.5 (raa+rbb) (rab)
–
(0.001) (0.001) (0.05) (0.05)
Age diff. (yrs)
Genetic similarity / year
2 2 1 1
0.46 0.62 0.73 0.73
1
0.63
Fig. 4 Proportion of queens surviving from year to year as a function of the average relatedness between newly adopted queens and existing queens. Dashed lines indicate two plausible levels of relatedness between newly adopted queens and existing queens : r = 0.4 is the empirically observed estimated relatedness for daughter gynes produced by nests; r = 0.2, half this amount, would be the relatedness value should colonies adopt, on average, 50 % daughters and 50 % non-relatives
(Fig. 4). This estimate applies only to the average reproductive longevity of queens. Queens might be physically present in nests of longer periods of time without reproducing, or they might reproduce at a reduced rate. Individual nests disappeared at a rate of 34% (n = 64) and 41 % (n = 66) per year in the Romar and Bear sites, respectively. Although colony migration probably explains a substantial fraction of these nest disappearances, it is also possible that colonies
281
Fig. 5 Cumulative fitness gained through direct and indirect means for an established queen whose last offspring are produced in year 0. Her indirect fitness, relative to that in her final year of life, is calculated for the 4 years following her death. Assumptions include an annual survival rate of 0.5 for other queens in the nest and genetic similarity between colony members from one year to the next of 0.6. Shown for colonies with average queen numbers of 1.5, 3, and 10
suffer high levels of mortality. Colonies that die are probably replaced, in most cases, by buds from existing nests. In M. tahoensis, a major component of a queen’s genetic contribution to future generations occurs after she has ceased laying eggs. From the standpoint of a queen in her last year of reproduction, she will gain half again as much inclusive fitness in the year after her death, if she can ensure that her colony continues without her (Fig. 5). Newly adopted queens can expect to receive, on average, a roughly equal fitness benefit through the reproduction of related nestmates after their deaths when compared to that received through the combination of direct and indirect reproduction while they are alive (Fig. 6). This ratio is relatively insensitive to the queen mortality rate, given the levels of genetic turnover found in M. tahoensis. In both analyses, queen number, per se, does not strongly affect the predicted immediate versus long-term fitness returns. Queen introductions Non-nestmate queens were routinely adopted into laboratory colonies. In 24 of 104 trials (23 %), entering queens were allowed to remain with the brood, and any antagonism toward these queens soon ended. There was no significant relationship between the queen number of host nests and their tendency to accept queens (Table 4). Among the 24 queens used in the transfers, some were better able than others to successfully invade established nests (repeated-measures analysis of variance, F22, 57 = 2.85, P < 0.001). Introduced queens that moved directly toward the host brood pile appeared to evade worker attacks most easily. Within the queen-
Fig. 6 Expected direct and indirect fitness while a queen is alive, as a proportion of her combined fitness while living and after her death. Fitness is shown as a function of average queen number (n = 1.5, 3, 10) and the proportion of queens surviving year-to-year (ranging from 0 to 0.6). Both direct and indirect fitness were calculated over five seasons, as it is unlikely that queens will affect colony fitness in a substantial way for longer than 5 years Table 4 Rates of acceptance and rejection of non-nestmate queens by queenless, monogynous, and polygynous colonies. There was no significant effect of queen number on the tendency of nests to accept queens (repeated-measures analysis of variance, F2,29 = 1.8, NS) Host colony status
Accepted
Rejected
No queen Monogynous Polygynous Total
12 3 9 24
43 18 19 80
number classes, particular nests also differed in their tendency to adopt queens (F22, 57 = 3.92, P < 0.001). Queens were always accepted when returned to their native nests.
Discussion Like other species in the genus (Elmes 1980), Myrmica tahoensis shows colony-level variation in queen number. Supernumerary queens are fully fertile, and as a consequence there is substantial colony-to-colony variation in the genetic relatedness of nestmates. Demographic data suggest that queens only rarely initiate colonies independently. Instead, queens are adopted into established nests, giving these nests lifespans that are much longer, on average, than those of any individual queen. The reproductive success of M. tahoensis queens is therefore determined mainly by their abilities to enter or return to established nests, not by their abilities to establish new colonies. Relatedness levels among adult workers and worker pupae imply effective queen numbers that are very close to those seen, on average, in M. tahoensis nest
282
excavations. These results suggest that queens produce workers at roughly equal rates, a finding that is corroborated by behavioral observations and by the fact that queens in laboratory colonies lay eggs at roughly equal rates (Evans 1996). Curiously, larval relatedness levels suggested higher queen numbers than did relatedness among adults, although this result only closely approached statistical significance. Further sampling, especially of larvae and sexual offspring, will help determine whether low levels of reproductive skew occur in colonies of M. tahoensis. Nevertheless, skew does not appear to be a strong component of the M. tahoensis breeding system. The excess homozygosity inferred from individual genotypes suggests that inbreeding is a component of the M. tahoensis mating system. Inbreeding occurs only rarely in ants (reviewed by Crozier and Pamilo 1996). In subalpine populations of M. tahoensis, excess homozygosity may arise from at least two mating-system phenomena. First, relatives might mate within nests. While high levels of nestmate relatedness suggest that such matings would result in substantial inbreeding, M. tahoensis colonies tend to specialize on the production of one sex or the other (Evans 1995). Therefore, this explanation would not apply to over two-thirds of reproductive colonies (and nearly half of those colonies that produce female sexuals). Localized mating between neighboring M. tahoensis colonies within a meadow could also result in excess homozygosity, when individual genotypes are compared to allele frequencies for the entire meadow (the Wahlund effect; Hartl and Clark 1989). Within-meadow genetic structure of the sort that might lead to excess homozygosity has not yet been observed in M. tahoensis (J.D, Evans, unpublished work), thus the occurrence of non-zero Fis levels in these populations remains an enigma. Several lines of evidence suggest that queens in colonies of M. tahoensis have short adult lifespans. First, worker-worker relatedness in individual colonies is poorly correlated with queen number. This result implies that queen number often changes between egglaying and the eclosion of adult workers. More directly, the pattern of relatedness among colony members of different ages suggests yearly rates of genetic turnover of about 40%. Together with estimates of the relatedness of colony members to their adopted queens, this turnover suggests that more than half of a colony’s queens die, or leave their nest, annually. Relatively short queen lifespans have also been described in Myrmica ruginodis and M. lobicornis (Seppä 1994), and inferred from demographic data for several other Myrmica species (Elmes 1980). Since Myrmica colonies are likely to reproduce by budding (reviewed by Bourke and Franks 1995, pp. 339–340), it is likely that some fraction of queen turnover in colonies reflects the dispersal of existing queens into newly budded colonies. While the rate of colony budding remains unknown, three meadow populations of M. tahoensis showed no signs
of the spatial-genetic structure expected when budding is a frequent event (J.D. Evans, unpublished work). To fully explain the genetic turnover data presented here, colonies would be required to produce autonomous buds roughly once per year, a rate that seems unlikely given the lack of local genetic structure in this species. Nevertheless, the occurrence of local genetic structure in populations of Myrmica ruginodis and M. rubra (Seppä and Pamilo 1995) suggests that budding (or temporary polydomy) might be frequent these species. Further analyses of both genetic structure and nest demography are required to assess the importance of budding in explaining the high levels of queen turnover found for Myrmica ants. Given substantial queen mortality, M. tahoensis colony members must decide often whether to accept new queens into their nest. Queen adoption decisions might depend on the actions of entering queens, established queens, and the workers of established nests. Since approaching queens are almost certain to increase their fitness by entering established nests, critical queenadoption decisions most likely rest on the actions of a colony’s existing queen and /or workers. It is difficult to determine whether established workers or queens effectively control queen adoption in the field. In laboratory colonies, M. tahoensis workers, but not queens, are aggressive toward unrelated queens that seek entrance into nests, suggesting that workers may limit queen adoption more than queens do (Evans 1996). However, under many sets of assumptions, workers are expected to be more willing than established queens are to see their nest adopt a new queen; this is particularly true when the potential adoptee is a daughter of the nest she seeks to enter (Nonacs 1988). In M. tahoensis, for example, workers are almost exclusively full sisters with the gynes produced by their colony, while queens tend to be more distantly related to these gynes (Evans 1995). Existing queens might avoid aggression toward new queens, not because they support the adoption of these new queens, but because of the inherent risks involved in aggressive encounters. Of course, established queens could affect the success of newlyentered queens more subtly, by pheromonal control over fecundity (cf. Bourke 1993) or by eating the eggs of newly introduced queens (Evans 1996). High rates of queen mortality, combined with relatively long colony lifespans, suggest that M. tahoensis workers and queens can gain significant amounts of inclusive fitness by ensuring the continued survival of their colonies after they die. Colonies might increase their survivorship by adopting new queens, since adopted queens will decrease the probability that a colony becomes queenless in any given year. Queenadoption decisions should reflect the effects of new queens on colony survivorship and their relatedness to members of the adopting colony. These two components, in conjunction with colony sex-allocation patterns, can be used to predict the timing and extent
283
of queen adoption in secondarily polygynous ants. A comparison of observed and predicted rates of adoption might help to determine whether these rates best reflect the inclusive fitness preferences of existing queens or of their workers. High relatedness levels among nestmate queens suggest that M. tahoensis colony members tend to adopt related individuals as new queens. Nonetheless, laboratory colonies regularly adopted unrelated queens, in contrast with Leptothorax curvispinosus, where colonies almost never adopt alien queens (Stuart et al. 1993). Further, the coexistence of distantly related queens in some nests suggests that queen adoption is not strictly limited to daughters. At least two hypotheses could explain the adoption of unrelated queens into M. tahoensis colonies. First, colony members might be unable to accurately assess the relatedness of approaching queens, allowing unrelated queens to parasitize existing nests (Elmes 1973). However, the fact that M. tahoensis workers unanimously discriminate against non-nestmate workers suggests that they could do the same for queens. Second, workers (and nestmate queens) might benefit from the adoption of a new queen even when she is unrelated to them, given that she is expected to substantially increase colony productivity (Nonacs 1988) or survivorship. Answers to several important questions are required in order to more fully assess the long-term implications of queen adoption in M. tahoensis and other secondarily polygynous ants. Most importantly, the impact of adopted queens on colony survivorship and productivity is poorly understood. Additional queens are unlikely to affect instantaneous colony productivity substantially, since productivity in ant colonies is correlated, in general, not with queen number but with overall colony size (Herbers 1984; Vargo and Fletcher 1986). More realistically, adopted queens could increase colony survivorship by “rescuing” the queenless colonies that are common in populations of Myrmica (Elmes 1980) and other facultatively polygynous ants. In M. tahoensis, high queen mortality rates and low average queen numbers place colonies at particularly high risk of becoming queenless. A second major question for queen-adoption strategies of ants involves the relationship between sexallocation patterns and the potential to adopt related queens (Nonacs 1993). Recently, Heinze et al. (1995) described the covariance of sex-allocation patterns and genetic relatedness in colonies of the facultatively polygynous ant Leptothorax acervorum. Their genetic data suggest that L. acervorum colonies depend primarily on the recruitment of daughter queens to maintain their high levels of colony relatedness. In M. tahoensis, a minority of colonies produce female sexuals in a given year (Evans 1995). M. tahoensis colonies that do not produce female sexuals are quite likely to be constrained in their abilities to recruit related queens, placing a potentially heavy cost on the
sex-ratio specialization present in this species. A final important question, and one central to kin-selection arguments generally (e.g., Grafen, 1990) involves the abilities of nestmates to accurately assess relatedness (or surrogates of relatedness such as cues picked up from a common nest) while making queen-adoption decisions. An understanding of the accuracy with which colony members assess would-be queens might help to resolve whether distantly related queens are undetected parasites on colonies, or whether their adoption reflects an urgent need by colonies to replace their own dying queens. Acknowledgements K. Bright, K. Chu, C. Nufio and A. Valdez helped immensely with field collections and behavioral observations. J. Seger and M. Richards provided helpful advice on both the project and this paper. I. Billick, A. Bourke, K. Ross, P. Seppä, and two anonymous reviewers provided comments that greatly improved the manuscript. I thank A. Francoeur for his help in identifying specimens of Myrmica. Financial support was provided by a U.S. National Science Foundation Graduate Fellowship, by a Genetics Training Grant from the U.S. National Institutes of Health, by the Rocky Mountain Biological Laboratory, and by the University of Utah.
References Bourke AFG (1993) Lack of experimental evidence for pheromonal inhibition of reproduction among queens in the ant Leptothorax acervorum. Anim Behav 45 : 501–509 Bourke AFG, Franks NR (1995) Social evolution in ants. Princeton University Press, Princeton Brian MV (1983) Social insects: ecology and behavioural biology. Chapman and Hall, London Brian MV, Jones RM (1980) Worker population structure and gyne production in the ant Myrmica. Behav Ecol Sociobiol 7 : 281–286 Crozier RH, Pamilo P (1996) Evolution of social insect colonies: sex allocation and kin selection. Oxford University Press, Oxford Elmes GW (1973) Observations on the density of queens in natural colonies of Myrmica rubra L. (Hymenoptera : Formicidae). J Anim Ecol 42 : 761–771 Elmes GW (1980) Queen numbers in colonies of ants of the genus Myrmica. Insectes Soc 27 : 43– 60 Evans JD (1993) Parentage analyses in ant colonies using simple sequence repeat loci. Mol Ecol 2 : 393–397 Evans JD (1995) Relatedness threshold for the production of female sexuals in colonies of a polygynous ant, Myrmica tahoensis, as revealed by microsatellite DNA analysis, Proc Natl Acad Sci USA 92 : 6514 – 6517 Evans JD (1996) Competition and relatedness between queens of the facultatively polygynous ant Myrmica tahoensis. Anim Behav 51 : 831–840 Franks NR, Ireland B, Bourke AFG (1990) Conflicts, social economics and life history strategies in ants. Behav Ecol Sociobiol 27 : 175–181 Grafen A (1990) Do animals really recognize kin? Anim Behav 39 : 42–54 Hamilton WD (1972) Altruism and related phenomena, mainly in social insects. Annu Rev Ecol Syst 3 : 193–232 Hartl D, Clark AG (1989) Principles of population genetics, 2nd edn. Sinauer, Sunderland, Mass Heinze J, Lipski N, Schlehmeyer K, Hölldobler B (1995) Colony structure and reproduction in the ant, Leptothorax acervorum. Behav Ecol 6 : 359–367
284 Herbers JM (1984) Queen-worker conflict and eusocial evolution in a polygynous ant species. Evolution 38: 631–643 Herbers JM (1993) Ecological determinants of queen number in ants. In: Keller L (ed) Queen number and sociality in insects. Oxford University Press, Oxford, pp 262–293 Hölldobler B, Wilson EO (1990) The ants. Belknap Press, Cambridge Keller L (1993a) Queen number and sociality in insects. Oxford University Press, Oxford Keller L (1993b) The assessment of reproductive success of queens in ants and other social insects. Oikos 67:177–180 Keller L, Passera L (1989) Size and fat content of gynes in relation to the mode of colony founding in ants (Hymenoptera; Formicidae). Oecologia 80:236–240 Keller L, Reeve HK (1994) Partitioning of reproduction in animal societies. Trends Ecol Evol 9:98–102 Nonacs P (1988) Queen number in colonies of social Hymenoptera as a kin-selected adaptation. Evolution 42:566–580 Nonacs P (1993) The effects of polygyny and colony life history on optimal sex investment. In: Keller L (ed) Queen number and sociality in insects. Oxford University Press, Oxford, pp 110–131 Pamilo P (1991) Evolution of colony characteristics in social insects. II. Number of reproductive individuals. Am Nat 138:412–433 Pemberton JM, Slate J, Bancroft DR, Barrett A (1995) Non-amplifying alleles at microsatellite loci: a caution for parentage and population studies. Molec Ecol 4:249–252 Potvin C, Lechowicz MJ, Tardif S (1990) The statistical analysis of ecophysical response curves obtained from experiments involving repeated measures. Ecology 71: 1389–1400
Queller DC (1993) Genetic relatedness and its components in polygynous colonies of social insects. In: Keller L (ed) Queen number and sociality in Insects. Oxford University Press, Oxford, pp 132–135 Queller DC, Goodnight KF (1989) Estimating relatedness using genetic markers. Evolution 43 : 258–275 Ross KG (1993) The breeding system of the fire ant Solenopsis invicta, and its effects on colony genetic structure. Am Nat 141 : 554–576 Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning, a laboratory manual, 2nd edn. Cold Spring Harbor Laboratory, New York Seppä P (1994) Sociogenetic organization of the ants Myrmica ruginodis and Myrmica lobicornis: Number, relatedness and longevity of reproducing individuals. J Evol Biol 7 : 71–95 Seppä P, Pamilo P (1995) Gene flow and population viscosity in Myrmica ants. Heredity 74 : 200–209 Smith DR (1979) Family Formicidae. In : Krombein KV, Hurd PD Jr, Smith DR, Burks BD (eds) Catalog of Hymenoptera in America north of Mexico, vol 2. Smithsonian Institution Press, Washington, D.C., pp 1323–1467 Stuart RJ, Greshambissett L, Alloway TM (1993) Queen adoption in the polygynous and polydomous ant, Leptothorax curvispinosus. Behav Ecol 4 : 276–281 Vargo EL, Fletcher DJC (1986) Queen number and the production of sexuals in the fire ant, Solenopsis invicta (Hymenoptera : Formicidae). Behav Ecol Sociobiol 19 : 41– 47 Wright S (1951) The genetical structure of populations. Ann Eugen 15 : 323–354
Communicated by R.H. Crozier
