Behavioral Ecology Advance Access originally published online on December 13, 2006
Behavioral Ecology 2007 18(2):375-383; doi:10.1093/beheco/arl088
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Reproductive specialization in multiple-queen colonies of the ant Formica exsecta
Department of Ecology and Evolution, Biophore, University of Lausanne, CH-1015 Lausanne, Switzerland
Address correspondence to R. Kümmerli, who is now at Institute of Evolutionary Biology, University of Edinburgh, King’s Buildings, Edinburg EH9 3JT, UK. E-mail: rolf.kuemmerli{at}ed.ac.uk.
Received 19 May 2006; revised 8 November 2006; accepted 18 November 2006.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In polygynous (multiple queens per nest) colonies of social insects, queens can increase their reproductive share by laying more eggs or by increasing the proportion of eggs that develop into reproductive individuals instead of workers. We used polymorphic microsatellite loci to determine the genetically effective contribution of queens to the production of gynes (new queens), males, and 2 different cohorts of workers in a polygynous population of the ant Formica exsecta. For this purpose, we developed a new method that can be used for diploid and haplodiploid organisms to quantify the degree of reproductive specialization among breeders in societies where there are too many breeders to ascertain parentage. Using this method, we found a high degree of reproductive specialization among nest-mate queens in both female- and male-producing colonies (sex ratio is bimodally distributed in the study population). For example, a high effective proportion of queens (25% and 79%, respectively) were specialized in the production of males in female- and male-producing colonies. Our analyses further revealed that in female-producing colonies, significantly fewer queens contributed to gyne production than to worker production. Finally, we found significant changes in the identity of queens contributing to different cohorts of workers. Altogether, these data demonstrate that colonies of F. exsecta, and probably those of many other highly polygynous social insect species, are composed of reproductive individuals differing in their investment to gynes, males, and workers. These findings demonstrate a new aspect of the highly dynamic social organization of complex animal societies.
Key words: ants, breeding system, polygyny, reproductive skew, reproductive specialization, social insects.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
One of the major transitions in evolution is the shift from solitary organisms to societies with reproductive division of labor (Maynard Smith and Szathmary 1995
; Szathmary and Maynard Smith 1995). A key feature differentiating animal societies is the apportionment of reproduction among individuals (Keller and Reeve 1994). Societies where all individuals reproduce equally versus societies where a single individual completely monopolizes reproduction represent end points of a continuum in the variance in the reproductive output among group members. In high-skew societies, reproduction is limited to one or a small subset of breeders whereas in low-skew societies, reproduction is more evenly distributed among group members (Keller and Reeve 1994).
In social Hymenoptera, reproductive skew is generally very high in colonies containing a single queen because queens typically monopolize most of the reproduction (Bourke 1988; Jeon and Choe 2003; Hammond and Keller 2004). However, in many species, colonies may contain 2 or more fertile queens (Keller 1993b, 1995; Bourke and Franks 1995). In such polygynous colonies, there is considerable variation between queens in their relative contribution to reproduction. Most studies conducted to date investigated the extent to which nest-mate queens differ in their overall reproductive output (reviewed in Reeve and Keller 2001). However, even if queens contribute similar numbers of eggs, there can be considerable variation in reproductive success (Keller 1993a) because the relative reproductive success of nest-mate queens does not depend only on their fecundity but also on the type of egg produced (haploid male or diploid female) and their developmental fate (development of diploid eggs into queens or workers).
Only few studies have compared the relative contribution of queens to the production of gynes (new queens), males, and/or workers under natural conditions. In the ants Solenopsis invicta and Formica sanguinea, the effective number of queens contributing to gyne production is lower than the effective number of queens contributing to worker production (Ross 1993; Pamilo and Seppä 1994). In the ant Pheidole pallidula, a significant trade-off has been found between a queen's contribution to gyne and worker offspring but not between a queen's investment in gyne versus male production (Fournier et al. 2004). Finally in Leptothorax acervorum, there is evidence for a trade-off occurring between a queen's contribution to male and worker production (Hammond et al. 2006) as well as to male and gyne production (Bourke et al. 1997) but not between a queen's investment in gyne versus worker production.
A few more studies have investigated the differential contribution to castes in artificially composed colonies in the laboratory. In the fire ant S. invicta, reproductive skew is higher for gyne than for worker production (Ross 1988). A significant trade-off between a queen's contribution to male and gyne production has been demonstrated in Linepithema humile (Fournier and Keller 2001). In Pachycondyla inversa, reproductive skew was low but the data did not allow comparing the degree of reproductive skew for the production of all castes (Heinze et al. 2001). Finally, in Leptothorax rugulatus, queens were significantly specialized in the production of a particular class of individuals (gynes, males, or workers) or a combination of classes in 8 out of 23 colonies (Rüppell et al. 2002).
Altogether, these data show that queens may frequently contribute differently to the production of gynes, workers, and males. However, the large majority of these studies have been conducted on species that host only few queens per colony [2–7, but see Ross (1993) for an exception] and analyses were often limited to 2 classes of individuals. Obtaining data on reproductive specialization is particularly challenging for colonies containing many queens because it is very difficult to ascertain maternity when it is not possible to collect all queens. Examinations of such highly polygynous social systems, however, are of special interest because the range of possible variation in reproductive partitioning increases with the number of breeders, and detailed analyses of specialization among queens could provide important insights into the organization of these insect societies.
The aim of this study was to develop a general method that quantifies the degree of reproductive specialization among breeders in societies where many individuals contribute to offspring production. Importantly, this method can be used for haplodiploid (males are haploid and females diploid) as well as diploid organisms. It is based on relatedness estimates within and between classes of offspring (sexes, castes, cohorts, etc.) and estimates the genetically effective proportion of breeders contributing to a single class or a combination of classes. We applied this method to polygynous colonies of the ant Formica exsecta in a population of the Swiss Jura Mountains. In this population, colonies contain many queens (Cherix et al. 1980; Brown and Keller 2000, 2002; Liautard and Keller 2001), and colony sex ratio is bimodally distributed (Brown and Keller 2000) with most colonies producing exclusively males and workers (referred to as male-producing colonies) and some colonies producing gynes, males, and workers (referred to as female-producing colonies). In F. exsecta, no brood overwinters and thus brood eclosing from July to October derives from eggs laid earlier that same year (Brown and Keller 2002). In July, sexuals (gynes and males) eclose along with the first cohort of workers (referred to as "summer workers"). Generally, the sexuals engage in mating activities that are followed by the recruitment of newly mated queens back to the parental colony (Brown and Keller 2002). After the release of sexuals in July, sexual production ceases, whereas worker production continues until late September/early October (Kümmerli R, Keller L, unpublished data) with these workers being referred to as "fall workers." We collected brood (gynes, males, summer, and fall workers) from male- and female-producing colonies and estimated the relatedness within and between classes of individuals using polymorphic microsatellite loci. We first used these relatedness estimates to compare the genetically effective number of queens contributing to each class of offspring and then applied our new method to estimate the degree of reproductive specialization among queens.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Sampling
The study was carried out at "les Chenevières," a cattle pasture at 1220 m in the Swiss Jura Mountains. In this population, colony density is very high (Liautard et al. 2003
), and colony sex ratio is strongly male biased with about 90% of the nests producing only males (Brown and Keller 2000). In 2002, we marked 411 colonies individually. On average, we collected 50 pupae per colony in July 2002 prior to the mating flight (summer brood). We determined the sex and the caste of each individual (gyne, male, or worker) by moistening pupae in ethanol and observing the size, eye shape, and genitalia of late-stage pupa (Brown and Keller 2000). Out of the 411 colonies, 20 male- and 16 female-producing colonies were randomly selected, and for each of these focal colonies, pupae of each class were prepared for genetic analysis and stored at –20 °C (gyne, worker, and male pupae for female-producing colonies and worker and male pupae for male-producing colonies). At the end of the breeding season (late September/early October), when only workers are produced, we collected pupae from all focal nests still producing offspring (15 female- and 17 male-producing nests). Overall, we genotyped 9.8 ± 3.8 gyne pupae [mean ± standard deviation (SD)], 16.4 ± 4.6 male pupae, 15.0 ± 5.2 summer worker pupae, and 15.6 ± 1.6 fall worker pupae per colony.
Laboratory methods
Genotypes of gyne, worker, and male pupae were determined at 8 microsatellite loci FL21 (Chapuisat 1996) and FE17, FE19, FE21, FE37, FE42, FE49, and FE51 (Gyllenstrand et al. 2002). DNA was extracted from the entire pupae in 500 µl of 5% Chelex and incubated for 10 min at 90 °C, vortexed, and incubated for another 10 min at 90 °C. Three multiplex polymerase chain reaction (PCR) were used: 1) multiplex PCR1 coamplified FE19, FE21, and FE51; 2) multiplex PCR2 coamplified FE17 and FE37; and 3) multiplex PCR3 coamplified FE42 and FE49. The locus FL21 was amplified in a simplex PCR. Amplification reactions were carried out in 10 µl reaction volumes. Full details of the amplification conditions for multiplex and simplex PCR are given in Kümmerli et al. (2005). PCR products were mixed and run on an automatic sequencer (ABI Prism 377XL). The number of alleles per locus ranged from 5 to 26 (11.6 ± 7.8, mean ± SD) with expected heterozygosities ranging between 0.44 and 0.84. Genotypes at the locus FE49 deviated significantly from Hardy–Weinberg equilibrium with the observed heterozygosity being much lower than the expected heterozygosity. This suggests that one or more null alleles were present at this locus. For this reason, we excluded FE49 from further analyses.
Comparing genetically effective number of breeders
A first test of reproductive specialization can be obtained by comparing whether the genetically effective number of breeders contributing to different offspring classes differs from one another. The genetically effective number of breeders, which corresponds to the number of equally reproducing breeders required to account for an observed relatedness pattern among offspring (Queller 1993; Ross 1993), can be calculated for diploid classes (Ne,dc) of offspring as
 | (1) |
Similarly, the genetically effective number of breeders contributing to haploid offspring classes (Ne,hc) is given by
 | (2) |
, 2001) and Queller (1993).
We applied these equations to polygynous F. exsecta colonies by estimating rdc for gyne, summer, and fall worker offspring and rhc for male offspring with the program RELATEDNESS 5.0.8 (Queller and Goodnight 1989). We obtained standard errors (SEs) by jackknifing over nests. We further used relatedness values for rfb, rmb, and me that have been previously estimated in our study population to be 0.084 ± 0.006 (mean ± SE), 0.020 ± 0.003 (mean ± SE) and 1.2, respectively (Haag-Liautard C, Pedersen JS, Ovaskainen O, Keller L, unpublished data).
Measuring reproductive specialization among queens
To quantify the reproductive specialization among breeders in their contribution to different classes of offspring, we used the equation developed by Pedersen and Boomsma (1999), which is analogous to the classic Jaccard index of similarity (Chao et al. 2005). This equation was initially used to estimate the genetically effective turnover of breeders in colonies of social insects across years, hence estimating the genetically effective proportion of breeders contributing to only one of 2 age cohorts. Thus, if this formula is applied to different classes of offspring that are produced simultaneously, it estimates the genetically effective proportion of breeders (
c) contributing to only one of 2 offspring classes. Hence, it provides a measure of reproductive specialization among breeders given by
 | (3) |
For comparisons of 2 classes of offspring having the same ploidy level, the genetically effective number of breeders can be substituted by the corresponding relatedness values and becomes
 | (4) |
c2 is the symmetrical relatedness between individuals of class 1 and class 2. Equations 3 and 4 are generalized versions of equations previously presented by Pedersen and Boomsma (1999).
We calculated the relatedness among gynes, summer, and fall workers as well as the relatedness between gynes and summer workers, gynes and fall workers, and summer and fall workers. We then used these relatedness estimates to apply Equation 4 in 3 pairwise comparisons to calculate the genetically effective proportion of queens (c) contributing to only one of 2 offspring classes under comparison. Finally, we calculated the SE for c by jackknifing over colonies.
Because the calculation of c involves 3 different relatedness values, each of them with some error, we tested the reliability of our estimates by conducting simulations. First, we obtained SEs for all relatedness values by jackknifing over loci. Second, we simulated the sensitivity of c with respect to the SE in the relatedness estimates. To do so, we considered deviations within the range of the relatedness SE that would lead to lower c values. We started from the observed c assuming no error and then added errors for all 3 estimates in steps of 5% of their total SE until the lowest possible estimate of c was reached (which was the case for rc1 – SE, rc2 – SE, and rc1 c2 + SE). For each of the 20 steps, we calculated c and tested whether it was significantly greater than zero by applying 1-tailed t-tests and using the SE obtained by jackknifing over colonies (see above).
The estimates of c can further be used to determine the genetically effective number of breeders being specialized in the production of offspring class 1 (Ne,c1,only), offspring class 2 (Ne,c2,only), or both offspring classes (Ne,c1 + c2) using the following formulas
 | (5) |
 | (6) |
 | (7) |
We conducted 3 pairwise comparisons to estimate the genetically effective number of queens being specialized in gyne versus summer worker production, gyne versus fall worker production, and summer versus fall worker production.
When combining all 3 pairwise comparisons, there were 7 categories of queens, and we estimated their relative proportions under the assumption that they all had the same fecundity. The first 3 categories were those containing queens contributing to a single class of offspring (gynes, summer workers, or fall workers). The next 3 categories included queens contributing equally to the production of 2 offspring classes (i.e., gynes and summer workers, gynes and fall workers, or summer and fall workers). Finally, the last category of queens was composed of those contributing equally to all 3 offspring classes. The relative proportions of these 7 categories of queens provide a measure of reproductive specialization. A complete lack of specialization would be indicated by the presence of a single type of queen contributing equally to the production of all 3 offspring classes. At the other end of the continuum, a complete specialization would occur if there were 3 types of queen, each contributing to the production of a single offspring class. Intermediate levels of specialization would be reflected by different proportions of queens contributing to one or more classes.
For comparisons of 2 classes of offspring having different ploidy levels, Equation 4 can unfortunately not be used to estimate the genetically effective proportion of female breeders (only females can produce haploid offspring) being specialized in the production of diploid versus haploid offspring classes. This is because the relatedness between diploid and haploid offspring classes is asymmetric (rdc 
hc 
rhc dc). We therefore developed an interpolative approach where we calculated the expected relatedness of diploid toward haploid offspring classes (rdc hc) for given c values. Next we estimated the c value that led to the best match between the expected and observed relatedness values. The expected rdc hc, assuming that all female breeders (N) in a group contribute equally to diploid and haploid offspring, can be broken down into the sum of 2 terms specifying how these individuals are related through their mothers (first term) and through the average relatedness among female breeders (second term):
 | (8) |
hs is the relatedness of diploid siblings toward their haploid siblings and equals 1/4, while rfb is the relatedness among female breeders. Equation 8 is analogous to equations previously presented by Ross (1993, 2001) and Queller (1993) for relatedness among female and among male brood. We further processed this formula by assuming that there is reproductive specialization among breeders such that only a given effective proportion of female breeders (1 – c) contribute equally to diploid and haploid offspring. Under this scenario, the contribution of the first term to rdc hc decreases, whereas the contribution of the second term to rdc hc remains unaffected such that the expected rdc hc becomes
 | (9) |
 | (10) |
c. A summary of the different types of genetically effective breeder numbers used in this study is given in Table 1.
|
We used this method to conduct pairwise comparisons of reproductive specialization among queens for the production of males and gynes as well as for the production of males and summer workers. This method permits one to obtain an overall quantification of the differential investments of queens into the 2 offspring classes.
Statistical analyses
In colonies hosting many queens, the relatedness among brood is low and the estimated relatedness coefficients can sometimes be negative due to stochastic sampling errors arising from the limited number of loci and pupae genotyped. Such negative values lead to large negative estimates of effective queen number. To cope with this problem, we applied the reciprocal transformation 1/X to all estimates of effective queen number (Ne = X) prior to statistical analyses (Sokal and Rohlf 1995). This transformation yields a linear relationship between relatedness values and the reciprocals of the effective queen numbers, with the transformed values ranging from around zero (low or slightly negative relatedness values) to one (high relatedness values). Transformed values did not deviate from a normal distribution for summer brood (Shapiro–Wilk normality tests: all P > 0.2) but did so for fall workers (Shapiro–Wilk normality tests: all P < 0.05). Hence, to test whether the effective queen number contributing to offspring classes produced in summer differed within colonies, we applied a parametric repeated measures analysis of variance (ANOVA) for female-producing colonies (followed by Tukey post hoc comparisons) and a paired t-test for male-producing colonies. In contrast, we conducted pairwise nonparametric Wilcoxon matched-pairs signed-rank tests for the comparison between summer offspring classes and fall workers. Because we conducted multiple comparisons on the same data set, we applied the Bonferroni correction with the nominal 
= 5% being 0.0125 and 0.017 for female- and male-producing colonies, respectively.
To test whether there was a significant reproductive specialization among queens within a colony in their contribution to 2 classes of diploid offspring, we tested whether rc1 c2 (relatedness between classes) was significantly lower than rc1 + rc2 – rc1 c2 (relatedness within classes) using paired 1-tailed t-tests. The use of 1-tailed t-tests is justified because relatedness between classes can only be equal or smaller but never be greater than relatedness within classes. All relatedness values were normalized by applying logarithmic transformations prior to analyses. Due to multiple comparisons in female-producing colonies, the Bonferroni corrected probability for nominal = 5% was 0.017.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Comparison of genetically effective queen number
Relatedness estimates among members of the same offspring class were generally low (range 0.06–0.25, Table 2) but all significantly greater than zero (1-tailed t-tests: all P < 0.05). We used these relatedness estimates to calculate the genetically effective queen number contributing to the different offspring classes. In female-producing colonies, there were significant differences in the genetically effective number of queens contributing to gynes, males, and summer workers (Figure 1a, repeated measures ANOVA: F2,15 = 5.20, P = 0.012). A posteriori multiple comparisons revealed that significantly fewer queens effectively contributed to gyne brood than to male (Tukey test: P < 0.025) and worker brood (Tukey test: P < 0.05), while the genetically effective queen number producing males and workers did not differ significantly from one another (Tukey test: P > 0.05). Furthermore, the genetically effective number of queens contributing to fall workers was not significantly different from the genetically effective number of queens contributing to each of the 3 offspring classes produced in summer [Wilcoxon matched-pairs signed-rank test (n = 15): gynes P = 0.16, males P = 0.33, workers P = 0.91, Figure 1a].
|
|
In male-producing colonies, the genetically effective number of queens contributing to male production did not differ from the genetically effective number of queens contributing to summer workers (paired t-test: t19 = 1.90, P = 0.072). These 2 values were also not significantly different from the genetically effective number of queens contributing to fall workers [Wilcoxon matched-pairs signed-rank test (n = 17): males P = 0.52, workers P = 0.19, Figure 1b].
Queen specialization pattern
In female-producing colonies, gyne and worker pupuae were significantly more related to individuals of their own class than to individuals of the other class (1-tailed t-test: gyne–summer worker, t15 = 2.73, P = 0.008; gyne–fall worker, t14 = 3.21, P = 0.003, Table 3). In both, female- and male-producing colonies, summer and fall workers were more related to workers of their own class than those of the other class. This difference was however only significant in male-producing colonies (1-tailed t-test: t16 = 3.31, P = 0.002) but not in female-producing colonies (1-tailed t-test: t14 = 1.38, P = 0.09). In line with the finding that relatedness among female pupae tended to be higher within than between offspring classes, pairwise comparisons revealed that there was a high genetically effective proportion of queens (c) contributing to a single female offspring class (Table 3). Our simulations show that an error accumulation of the 3 relatedness values lead to a decrease of c estimates for all comparisons (Figure 2). Importantly, the c estimates for the gynes–summer workers and the gynes–fall workers comparison in female-producing colonies remained significantly positive for all simulated steps (1-tailed t-tests: all P < 0.05). For the summer–fall workers comparison, t-tests for observed and simulated c values also revealed consistent results by being all not significantly different from zero (one-tailed t-tests: all P > 0.05). In male-producing colonies, the simulated c values for the summer–fall workers comparison remained significantly positive for 19 out of the 20 simulated steps and only became insignificant when accumulating the maximal error for all 3 relatedness estimates (Figure 2). Altogether, these simulations demonstrate that our method to calculate the genetically effective proportion of queens produces reliable results and therefore provides a robust test to detect significant reproductive specialization among nest-mate queens.
|
|
For all pairwise comparisons, we calculated the genetically effective queen number that contributed to either one or both female offspring classes under comparison (Table 4). In female-producing colonies, the gyne–summer worker comparison revealed that a high percentage (75%) of the queens were effectively specialized in worker production, whereas only 25% of the queens effectively contributed to both gyne and worker production. The negative value of the estimated effective number of queens contributing to only gynes is probably a sampling error, and the data are compatible with no or only a few queens being effectively specialized in gyne production. The gyne–fall worker comparison in female-producing colonies was consistent with most queens effectively contributing to only fall workers (61%) and a lower proportion of queens effectively producing either gynes only (17%) or gynes and fall workers (22%). The comparison of the 2 worker cohorts indicated that most queens effectively contributed either to the summer workers (49%) or to both worker classes (47%) and that there were only few queens contributing exclusively to fall workers (4%). Finally, in male-producing colonies, the comparison between the 2 worker classes revealed that most queens effectively contributed to either fall workers only (49%) or both worker classes (37%). Only few queens effectively contributed to the production of only summer workers (14%).
|
The combination of the results of all pairwise comparisons in female-producing colonies allowed us to obtain an overall estimate of the relative contribution of queens to the summer and fall brood. Such a combination can be obtained by logical deduction, using the genetically effective queen numbers presented in Table 4, starting with the lowest estimate of Ne,c1 + c2 across all pairwise comparisons (Ne,c1 + c2 = 1.7 for the gynes–fall workers comparison, Table 4). This estimate shows that 1.7 queens effectively contributed to gynes and fall workers, although it is yet unknown whether these queens also produced summer workers. The gynes–summer workers comparison, however, reveals that no queens effectively contributed to gynes without also producing summer workers. Hence, the effective queen number that contributed to all 3 female offspring classes is equal to 1.7, whereas the effective queen number that contributed to gynes and summer workers but not to fall workers becomes 1.2 (=2.9 – 1.7). Accordingly, the genetically effective queen number contributing to summer and fall workers but not to gynes becomes 4.0 (=5.7 – 1.7), whereas the genetically effective queen number contributing to gynes and fall workers but not to summer workers becomes 0 (=1.7 – 1.7). Finally, the genetically effective queen number producing only a single offspring class are equal to 0 (=2.9 – 1.7 – 1.2) for gyne production, 4.8 (=11.7 – 1.7 – 1.2 – 4.0) for summer worker production, and 0.5 (=6.2 – 1.7 – 4.0) for fall worker production. Because there were no queens that effectively contributed only to gynes, our analysis provides one single solution of how the 3 pairwise comparisons can be combined. According to this combination, the best fit to our data was obtained when assuming that 39% of the queens contributed to summer workers only, 4% to fall workers only, 33% to both worker cohorts but not to gynes, 10% to gynes and summer workers but not to fall workers, and 14% to gynes and the 2 worker cohorts (Figure 3).
|
Our interpolative approach allowed us to estimate the extent to which queens were specialized in male versus female production. In female- and male-producing colonies, we found that the mean genetically effective number of queens contributing to male production (Ne,m) was higher than the genetically effective number of queens contributing to female production (Ne,f). Accordingly, this sets a minimal degree of specialization (
c – minimal) being equal to 1 – Ne,f/Ne,m (Table 5). Our analyses revealed that the best fit between the expected and observed relatedness values was obtained when c was closest to c – minimal. However, in all comparisons, the observed relatedness estimates tended to be slightly higher than the expected relatedness values even for c – minimal (Table 5).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Our genetic analyses revealed that, in female-producing nests, the effective number of queens (11.7) contributing to worker production in summer was about 4 times higher than the effective number of queens (2.9) contributing to gyne production (Table 4). Our analysis on the genetically effective proportion of queens (
c) contributing to the production of only workers or gynes showed that the observed pattern of genetic relatedness values was best explained by queens contributing to gyne production being a subset of those contributing to worker production (Table 4). Importantly, the estimate that about one-quarter of the queens contributed to the production of both gynes and workers is based on the assumption that all queens have the same fecundity and that, when contributing to several offspring classes, queens contribute equally to the production of these classes. If these assumptions were violated for this and all other analyses in our study, this would affect the estimated number of specialized queens (Table 4) but not alter the general conclusion of our study that a high proportion of queens are reproductively specialized. This is because all estimates of the proportion of effectively specialized queens (c) are very high (Tables 3 and 5), a pattern that cannot simply arise from queens differing in their fecundity.
In female- and male-producing colonies, the effective number of queens (11.7 and 6.8, respectively) contributing to summer workers was not significantly different than the effective number of queens (6.2 and 11.6, respectively) contributing to fall workers (Figure 1). However, our analysis on the genetically effective proportion of specialized queens (c) revealed that a high proportion of queens (0.53 and 0.63, respectively) contributed only to the production of summer or fall workers in female- and male-producing colonies (Table 3). This high degree of reproductive specialization can be explained either by some queens preferentially reproducing early or late in the reproductive season or by queen turnover (Pedersen and Boomsma 1999). In polygynous ants, queen life span is often low and queen turnover high (Keller and Genoud 1997). Hence, the pattern that different queens contributed to summer and fall workers can partly be explained by some queens having died and/or some new queens being recruited. It is also possible that some queens changed nests during summer, further accounting for the lower relatedness between the workers produced in summer and fall compared with the relatedness among workers of the same cohort. Distinguishing between these different hypotheses would require marking queens and estimating queen mortality and internest movement between summer and fall.
In female-producing nests, the effective number of queens contributing to male production (15.5) was about 5 times higher than the effective number of queens (2.9) contributing to gyne production (Table 5). Similarly, in both female- and male-producing colonies, there was a tendency for the effective number of queens (15.5 and 31.8, respectively) contributing to males to be greater than the effective number of queens (11.7 and 6.8, respectively) contributing to summer workers (Table 5). Our interpolative method estimating the genetically effective proportion of queens (c) specialized in male or female (gynes of workers) production revealed that our data are best explained by queens contributing to gyne or worker production being a subset of those contributing to male production. The pattern of fewer queens contributing to gyne and worker than to male production cannot simply be explained by the presence of noninseminated queens. This is because a previous study in the same population revealed that the proportion of noninseminated queens is 11% (Haag-Liautard C, Pedersen JS, Ovaskainen O, Keller L, unpublished data), whereas our estimated effective proportion of queens being specialized in male production is much higher (25–81%, Table 5). This indicates that, as well as noninseminated queens that can only produce males, a considerable proportion of inseminated queens were specialized in male production. Moreover, the estimated effective number of queens contributing to male production is unlikely to be overestimated due to worker reproduction because both field data from a monogynous population of the same species (Sundström et al. 1996) and our own unpublished data from a laboratory brood rearing experiment of workers from the same Swiss Jura mountain population indicate that worker reproduction is absent in queenright colonies.
Because previously there were no methods that allowed the quantification of queen specialization in colonies with high queen number, there are no comparable studies conducted under natural conditions in other species of social insects. However, information from natural colonies on comparisons of genetically effective queen number is available from 4 other ant species. In S. invicta and F. sanguinea, the genetically effective number of queen contributing to gynes is also lower than the effective number of queens contributing to workers (Ross 1993; Pamilo and Seppä 1994). By contrast, there was no significant difference in the effective number of queens contributing to gyne and male production in L. acervorum (Bourke et al. 1997). The only study that compared the contribution of queens to gyne, male, and worker production in the field was conducted on the ant P. pallidula (Fournier et al. 2004), where significantly fewer queens contributed to male than to gyne and worker production. Furthermore, differential contribution of queens to gyne and male production has also been observed in the hover wasp Liostenogaster flavolineata where gyne production was generally monopolized by a single queen, while several queens contributed to male production (Sumner et al. 2002). Altogether, these data reveal that nest-mate queens commonly differ in their relative contribution to different offspring classes under natural conditions, indicating that reproductive specialization seems to be a general characteristic of polygynous insect societies.
Our data showed that the genetically effective queen number across offspring classes and colonies was higher for male- (31.8) than for female-producing colonies (15.5, Table 5). The difference in queen number between male- and female-producing colonies is best explained by the queen-replenishment hypothesis positing that colonies containing many queens should only produce males, whereas colonies with few queens benefit from also producing and recruiting new queens to enhance colony survival and productivity (Brown and Keller 2000; Kümmerli et al. 2005). In relation to this, our results revealed that the pattern of reproductive specialization among queens was more extreme in male- than in female-producing colonies. First, the genetically effective proportion of queens being specialized in male production was higher in male- (79%) than in female-producing (25%) colonies. Second, a significant effective proportion of queens were specialized in the production of summer or fall workers in male- but not in female-producing colonies. This positive association between queen number and the degree of reproductive specialization among queens supports the view that reproductive partitioning gets more extreme with increasing number of breeders, a relationship that has already been suggested for the fire ant S. invicta (Ross 1993).
By combining data from this study and information on the biology of F. exsecta, we propose the following scenario to account for the pattern of queen specialization described in this study. Field observations revealed that in the studied population, the first adult workers, males, and gynes produced within a breeding season emerge simultaneously in July (Brown and Keller 2000; Liautard et al. 2003). Because workers are 1.5–3.4 times lighter than males or gynes (Brown and Keller 2000), they presumably need less time to develop from the egg to the adult stage. Thus, because no brood overwinters (Brown and Keller 2002), the first eggs produced end of March/beginning of April (personal observations) probably develop into gynes and males. Accordingly, the best explanation for our data is that all queens contributed to male production in early spring and that, in female-producing nests, a subset of these queens (19%) also produced diploid brood developing into queens. The summer workers most likely developed from the second batch of eggs laid in late April and May. These eggs were apparently laid by a subset of the queens that contributed to male production. This would be consistent with our findings that there were no queens contributing to summer workers but not to male production and that the effective number of queens contributing to summer workers (11.7 in female-producing nests and 6.8 in male-producing nests) was lower than the effective number of queens contributing to males (15.5 in female-producing nests and 31.8 in male-producing nests). Several factors may account for the lower number of queens contributing to workers than to males. The first is that 11% of the queens are unmated. Second, some mated queens may also specialize in the production of male eggs. Third, queen mortality in early spring may account for a decrease in the effective number of queens contributing to the second batch of eggs compared with the first batch of eggs. The fall workers developed from eggs that were laid after the emergence of the summer brood. Our data show that about half the queens contributing to summer workers were not the same as those contributing to fall workers. As mentioned earlier, this pattern could be explained by queens being reproductively specialized in the production of summer and/or fall brood or by an important queen turnover during the reproductive season.
In conclusion, this study presents a new method that quantifies the reproductive specialization among breeders in their contribution to different offspring classes in societies where there are too many breeders to assign parentage. We applied this method to highly polygynous field colonies of F. exsecta and could show that a high proportion of queens were apparently specialized in the production of only a subset of offspring classes present in colonies. Our data also revealed considerable changes in the identity of queens contributing to summer and fall workers. These data demonstrate that F. exsecta colonies, and probably those of many other highly polygynous social insect species, are composed of reproductive individuals which exhibit significant differences in their relative contributions to the different types of individuals composing the colonies. Moreover, the identity of these individuals, and perhaps their reproductive roles, vary considerably over time. Hence, our new method to quantify the degree of reproductive specialization should prove useful in future studies aiming at understanding the dynamic structure and complex social organizations characterizing many ants and other social organisms.
|
ACKNOWLEDGEMENTS |
|---|
We thank Rob Hammond, Michel Chapuisat, Jes Pedersen, Andrew Bourke, and 2 anonymous referees for constructive comments on the manuscript. We are grateful to Susanne Kümmerli-Schildknecht, Michael Ringli, and Grégoire Castella for their help in the field and in the laboratory. Many thanks to Cathy Haag-Liautard for providing unpublished data and useful information on the breeding system of the study species and to Tobias Suter who wrote a computer program that permitted to create Figure 3. This work was supported by several grants from the Swiss National Science Foundation.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bourke AFG. (1988) Worker reproduction in the higher eusocial Hymenoptera. Q Rev Biol 63:291–311.[CrossRef]
Bourke AFG and Franks NR. (1995) Social evolution in ants. (Princeton University Press, Princeton (NJ)).
Bourke AFG, Green HAA, Bruford MW. (1997) Parentage, reproductive skew and queen turnover in a multiple-queen ant analysed with microsatellites. Proc R Soc Lond B Biol Sci 264:277–283.[Medline]
Brown W and Keller L. (2000) Colony sex ratios vary with queen number but not relatedness asymmetry in the ant Formica exsecta. Proc R Soc Lond B Biol Sci 267:1751–1757.[Medline]
Brown WD and Keller L. (2002) Queen recruitment and split sex ratios in polygynous colonies of the wood ant Formica exsecta. Ecol Lett 5:102–109.[CrossRef]
Chao A, Chazdon RL, Colwell RK, Shen TJ. (2005) A new statistical approach for assessing similarity of species composition with incidence and abundance data. Ecol Lett 8:148–159.[Medline]
Chapuisat M. (1996) Characterization of microsatellite loci in Formica lugubris B and their variability in other ant species. Mol Ecol 5:599–601.[CrossRef][Medline]
Cherix D, Werner P, Catzeflis F. (1980) Organisation spatiale d'un système polycalique chez Formica (Coptoformica) exsecta Nyl. (Hymenoptera: Formicidae). Mitt Schweiz Entomol Ges 53:163–171.
Fournier D, Aron S, Keller L. (2004) Significant reproductive skew in the facultatively polygynous ant Pheidole pallidula. Mol Ecol 13:203–210.[CrossRef][Medline]
Fournier D and Keller L. (2001) Partitioning of reproduction among queens in the Argentine ant, Linepithema humile. Anim Behav 62:1039–1045.[CrossRef]
Gyllenstrand N, Gertsch PJ, Pamilo P. (2002) Polymorphic microsatellite DNA markers in the ant Formica exsecta. Mol Ecol Notes 2:67–69.[CrossRef]
Hammond RL, Bruford MW, Bourke AFG. (2006) A test of reproductive skew models in a field population of a multiple-queen ant. Behav Ecol Sociobiol 61:265–275.[CrossRef]
Hammond RL and Keller L. (2004) Conflict over male parentage in social insects. PLoS Biol 2:e248.
Heinze J, Trunzer B, Hölldobler B, Delabie JHC. (2001) Reproductive skew and queen relatedness in an ant with primary polygyny. Insectes Soc 48:149–153.[CrossRef]
Jeon J and Choe JC. (2003) Reproductive skew and the origin of sterile castes. Am Nat 161:206–224.
Keller L. (1993a) The assessment of reproductive success of queens in ants and other social insects. Oikos 67:177–180.[CrossRef]
Keller L. (1993b) Queen number and sociality in insects. (Oxford University Press, Oxford).
Keller L. (1995) Social life: the paradox of multiple-queen colonies. Trends Ecol Evol 10:355–360.
Keller L and Genoud M. (1997) Extraordinary lifespans in ants: a test of evolutionary theories of ageing. Nature 389:958–960.[CrossRef]
Keller L and Reeve HK. (1994) Partitioning of reproduction in animal societies. Trends Ecol Evol 9:98–102.[CrossRef]
Kümmerli R, Helms KR, Keller L. (2005) Experimental manipulation of queen number affects colony sex ratio investment in the highly polygynous ant Formica exsecta. Proc R Soc Lond B Biol Sci 272:1789–1794.[Medline]
Liautard C, Brown WD, Helms KR, Keller L. (2003) Temporal and spatial variations of gyne production in the ant Formica exsecta. Oecologia 136:558–564.[CrossRef][Web of Science][Medline]
Liautard C and Keller L. (2001) Restricted effective queen dispersal at a microgeographic scale in polygynous populations of the ant Formica exsecta. Evolution 55:2484–2492.[CrossRef][Web of Science][Medline]
Maynard Smith J and Szathmary E. (1995) The major transitions in evolution. (Freeman, Oxford).
Pamilo P and Seppä P. (1994) Reproductive competition and conflicts in colonies of the ant Formica sanguinea. Anim Behav 48:1201–1206.[CrossRef]
Pedersen JS and Boomsma JJ. (1999) Effect of habitat saturation on the number and turnover of queens in the polygynous ant, Myrmica sulcinodis. J Evol Biol 12:903–917.[CrossRef]
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. 132–152.
Queller DC and Goodnight KF. (1989) Estimating relatedness using genetic markers. Evolution 43:258–275.[CrossRef][Web of Science]
Reeve HK and Keller L. (2001) Tests of reproductive-skew models in social insects. Annu Rev Entomol 46:347–385.[CrossRef][Web of Science][Medline]
Ross KG. (1988) Differential reproduction in multiple-queen colonies of the fire ant Solenopsis invicta (Hymenoptera: Formicidae). Behav Ecol Sociobiol 23:341–355.[CrossRef]
Ross KG. (1993) The breeding system of the fire ant Solenopsis invicta, and its effects on colony genetic structure. Am Nat 141:554–576.[CrossRef]
Ross KG. (2001) Molecular ecology and social behaviour: analyses of breeding systems and genetic structure. Mol Ecol 10:265–284.[CrossRef][Medline]
Rüppell O, Heinze J, Hölldobler B. (2002) Intracolonial patterns of reproduction in the queen-size dimorphic ant Leptothorax rugatulus. Behav Ecol 13:239–247.
Sokal RR and Rohlf FJ. (1995) Biometry. (Freeman, New York (NY)).
Sumner S, Casiraghi M, Foster W, Field J. (2002) High reproductive skew in tropical hover wasps. Proc R Soc Lond B Biol Sci 269:179–186.[Medline]
Sundström L, Chapuisat M, Keller L. (1996) Conditional manipulation of sex ratios by ant workers: a test of kin selection theory. Science 274:993–995.
Szathmary E and Maynard Smith J. (1995) The major evolutionary transitions. Nature 374:227–231.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. Schwander and L. Keller Genetic Compatibility Affects Queen and Worker Caste Determination Science, October 24, 2008; 322(5901): 552 - 552. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Bargum and L. Sundstrom Multiple breeders, breeder shifts and inclusive fitness returns in an ant Proc R Soc B, June 22, 2007; 274(1617): 1547 - 1551. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||