Frequency of multiple paternity in Myrmica scabrinodis from southern Poland

In hymenopteran colonies, which comprise specialized infertile and sexual individuals, more than one male may contribute to the progeny of a single queen (polyandry). Polyandry is predicted to alter kin-selected conflicts over reproduction in haplodiploid insect societies (Keller & Reeve 1994; Boomsma & Ratnieks 1996; Strassmann 2001). Several direct and indirect benefits of multiple mating have been postulated, including selecting for or extracting higher queen mating frequencies (Hamilton 1987; Keller & Reeve 1994; Schmid-Hempel & Crozier 1999).

The occurrence of polyandry has been investigated in several ant taxa and also in Myrmica (Boomsma & Ratnieks 1996; Pedersen & Boomsma 1999; Strassmann 2001), which is an ant genus characterized by facultative polygyny (more than one queen can sire progeny in the colony). The frequency of multiple mating has been investigated for six Myrmica species: M. sulcinodis Nylander, M. ruginodis Nylander, M. rubra Linnaeus, M. tahoensis Wheeler, M. punctiventris Roger and M. lobicornis Nylander (Woyciechowski 1990; Boomsma & Ratnieks 1996; Pedersen & Boomsma 1999; and citations therein). In the present study we examined the level of polyandry in a population of M. scabrinodis Nylander by deducing the number of patrilines from genotype data for diploid offspring.

Material for this study was gathered in southern Poland in the Vistula valley, near Krakow city center (50°01′44″N, 19°51′52″E). An 8 ha plot was searched for Myrmica nests between the beginning of June and the beginning of July 2003. From each nest 50 adult workers were collected for species identification and for microsatellite analysis. In Myrmica populations, colonies are often queenless: at our study site, 12 of 30 excavated nests were found to be queenless in 2004 (P. Skórka, unpublished data). Moreover, the considerable queen turnover in this species (Seppä 1996) makes the functional queen difficult to identify. Therefore, the genotypes of the queens were deduced on the basis of the genotypes of the female offspring. Twenty-six M. scabrinodis nests were identified and 20 were randomly chosen for investigation of microsatellite polymorphisms. Total genomic DNA was extracted from 20 workers per nest. Worker genotypes were analyzed using three polymorphic microsatellite loci: Msca43, Msca64 and Msca78 (Henrich et al. 2003).

Analysis of the mating system within the population was performed using the MateSoft software package (Molainen et al. 2004), which is used to analyze codominant genetic marker data from male haplodiploid organisms. First, the microsatellite genotypes of workers were analyzed to check if individuals might be sired by one or more queens within particular nests. In further analyses only data from monogynous nests were used, as it was possible to deduce the correct patriline number only when this assumption was met. Next, the female offspring genotypes and deduced queen genotypes from monogynous colonies were used for deduction of genotypes of the putative males siring the offspring. Finally, the deduced data for queen and male genotypes were used in assigning offspring to patrilines corresponding to the queen's putative mates. These data were also used for the estimation of mating frequency statistics (Starr 1984; Molainen et al. 2004), that is, the average pedigree-effective mate number, m_e,p, and the proportion of double-mated queens among all mated queens, D. The first parameter gives information about the number of individual males siring the brood of a single queen in a population. The probability of erroneously assessing a double-mated queen as monandrous (a non-identification error; f′) due to either limited sample size or limited genetic resolution (Boomsma & Ratnieks 1996) was also calculated. For all estimations, 95% confidence limits and standard deviation values were estimated, respectively, by bootstrapping and jackknifing over colonies.

The numbers of alleles per locus were three for Msca43, seven for Msca64 and four for Msca78. The overall (not within-colony) gene diversity within loci (the heterozygosity expected; Nei 1987) was: 0.42 for Msca43, 0.39 for Msca64 and 0.39 for Msca78. Eleven colonies were categorized as functionally monogynous. Nine nests were rejected as being polygynous on the basis of information from one, two or all investigated loci. Deduction of male genotypes from deduced queen and female offspring data for the 11 colonies showed that five nests had one patriline, five nests had two patrilines and one nest had at least four patrilines. The observed proportion of multiply mated queens, D_obs, was 0.55 (SD = 0.05), which was lower than the corrected value (D_est = 0.71; SD = 0.06). The 95% confidence limits for both estimates (0.27–0.82 and 0.36–1.00, respectively) were wide, and this may be due to the high non-identification error, f′, of 0.23. This indicates that more queens need to be analyzed in order to obtain accurate estimates of D_est; however, doubly mated queens are unquestionably present in the population. Finally, the average pedigree effective mate number, m_e,p, was equal to 1.45 (SD = 0.06; 95% CI, 1.16–1.82). This value was clearly greater than one. For the nest with four patrilines, it is also possible to explain the variation in female offspring genotypes by having two doubly mated queens. Thus, alternative statistics for this scenario are as follows: for five nests with one patriline and seven nests with two patrilines, D_obs is equal to 0.50 (SD = 0.47; 95% CI, 0.25–0.75), and the corrected value, D_est, is equal to 0.69 (SD = 0.06; 95% CI, 0.34–1.00). Nonidentification error, f′, was 0.27. The average pedigree effective mate number, m_e,p, for the alternative scenario was 1.45 (SD = 0.06; 95% CI, 1.14–1.87).

Our findings indicate that multiple mating occurs within the investigated M. scabrinodis population. Double mating may occur with a frequency of 0.71, which is similar to frequencies obtained for M. rubra (D_exp = 0.65) and higher than frequencies obtained for other Myrmica species: 0.21 for M. ruginodis (Seppä 1994; Boomsma & Ratnieks 1996) and 0.40 for M. sulcinodis and M. tahoensis, and M. lobicornis and M. punctiventris in general mate only once (Pedersen & Boomsma 1999 and citations therein). Although multiple mating has been investigated for several ant taxa, their effective mate number is close to one, and double mating occurs at low frequencies (Boomsma & Ratnieks 1996; Strassmann 2001). The species studied here does not depart from this general trend.

Our observation that another Myrmica species mates more than once indicates that there are advantages to multiple mating. Among the many hypotheses that exist to explain this phenomenon, Alcock et al. (1978) proposed that multiple mating might have evolved as a male strategy that does not provide benefits to the queen or to colony functioning. This so-called convenience polyandry involves males maximizing the number of matings and the acceptance of additional female copulation events when it would be more costly for females to resist male sexual attempts than to mate. In facultatively polygynous species such as Myrmica, in large and dense populations with few vacant nest sites, sexuals may have a mixed mating strategy: some young females may stay and mate close to the natal nest and wait for readoption, and others may mate in swarms (large-scale mating flights) and disperse (Pedersen & Boomsma 1999 and citations therein). The cost of multiple mating is likely to be higher for dispersing females (including, for instance, the risk of predation) compared with the cost of mating for females that stay near the natal nest. Non-dispersing females may thus achieve a greater number of matings at a lower cost (Pedersen & Boomsma 1999). The analyses carried out within and across Myrmica and other facultatively polygynous ant species have shown that multiple mating, in the form of convenience polyandry, may be widespread (Pedersen & Boomsma 1999; Trontti et al. 2007).