SEARCH

SEARCH BY CITATION

Keywords:

  • Gp-9;
  • Pgm-3;
  • phenotype;
  • reproductive success;
  • selection;
  • Solenopsis invicta

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The objective of this study was to disentangle the relative effects of Pgm-3 and Gp-9 and/or other closely linked genes on the phenotypes and reproductive success of queens in introduced (USA) populations of S. invicta. Gp-9 or a closely linked gene(s) was found to have major effects on queen weight, the likelihood that queens shed their wings (a behaviour associated with the onset of reproduction), and the probability that queens are accepted in polygyne (multiple-queen) colonies. Our analyses show that once the effect of Gp-9 genotype is taken into account, Pgm-3 genotype no longer is significantly associated with differences in queen phenotype or the probability of queens being accepted in polygyne colonies. This suggests that the associations of Pgm-3 genotype with weight, wing shedding rate and probability of acceptance by polygyne colonies previously reported in studies that did not control for the effects of Gp-9 are due to the strong linkage disequilibrium that exists between Pgm-3 and Gp-9, or to linkage disequilibria between these and other genes affecting queen phenotype and fitness. Several lines of evidence, including data from the native South American range, suggest that additional cryptic alleles at Gp-9, or additional genes in the same linkage group as Gp-9, must be involved in controlling queen phenotype and the large suite of traits important in determining social organization of S. invicta colonies.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

An important goal in evolutionary biology is to understand the genetic basis of traits that influence the fitness of organisms in natural populations and, ultimately, to learn how the relationship of phenotype to genotype affects long-term evolution. Predicting such features as rates of phenotypic change and the constraints on selective responses requires knowledge of several key genetic parameters, most importantly, the number of loci underlying mean trait values and reaction norms, the distribution of effects among these loci and the genomic organization of such genes ( Barton & Turelli, 1989; Charlesworth, 1990; Clark & Koehn, 1991; Feder & Watt, 1991; Mitchell-Olds, 1995).

Studies of the genetic bases of selectively important traits involved in behavioural evolution have not been conducted frequently in natural populations ( Ferguson & Danzmann, 1985; Boake, 1994; Hoffmann, 1994; Lank et al., 1995 ; Shaw, 1996; Shuster & Sassaman, 1997). Data are particularly lacking for social organisms, where detailed empirical genetic analyses of the evolution of complex social behaviours have only recently begun to appear ( Moritz & Hillesheim, 1985; Page & Robinson, 1991, 1994; Keller & Ross, 1993a; Hunt et al., 1995 , 1998; Robinson & Page, 1995; Robinson et al., 1997 ; Robinson, in press ).

The fire ant Solenopsis invicta is one of the few social species in which genetic elements influencing both morphological and behavioural traits implicated in the expression of complex social organization have been identified ( Keller & Ross, 1993a, 1995, 1998; Ross & Keller, 1995; Ross et al., 1996a , b; Robinsonet al. 1997 ; Ross, 1997). Two distinctive social forms exist in this species, the monogyne form, in which colonies are headed by a single reproductive queen, and the polygyne form, in which colonies are headed by multiple reproductive queens. Variation at the enzyme-encoding gene phosphoglucomutase-3 (Pgm-3) has been shown to be associated with differences in the phenotype and reproductive success of queens in the polygyne form ( Ross, 1992; Keller & Ross, 1993a; Ross et al., 1996a ). Young polygyne queens with the homozygous genotype Pgm-3AA are prevented from becoming egg-layers in polygyne colonies because the workers invariably destroy such queens as they initiate reproduction ( Keller & Ross, 1993a). This differential survival and recruitment of new queens of different genotypes is associated with phenotypic differences ( Keller & Ross, 1993a, 1995); young, nonreproductive, winged queens with the genotypes Pgm-3Aa and Pgm-3aa weigh less, shed their wings less frequently (a behaviour associated with the onset of reproduction) and achieve lower fecundity than Pgm-3AA queens. In contrast, a queen’s genotype at this locus does not affect her phenotype or her probability of survival (or reproduction) in the monogyne form of this ant ( Ross, 1992; Keller & Ross, 1993a). Because phosphoglucomutase plays a central role in insect metabolism, it has been suggested that Pgm-3 directly influences queen phenotypes and is a target of selection in polygyne S. invicta ( Ross et al., 1996a ).

Another codominant mendelian locus, general protein-9 (Gp-9), recently has been shown also to be under strong selection in the polygyne form of S. invicta. All reproductive queens are heterozygous (Gp-9Bb) at this biallelic locus in polygyne colonies in the introduced (USA) range (The identity of the protein product of Gp-9 is unknown at present.). This is because (i) Gp-9bb apparently is an age-dependent lethal genotype in females ( Ross, 1997) and (ii) all Gp-9BB queens that attempt to initiate reproduction in polygyne nests are selectively executed by workers ( Keller & Ross, 1998). The strong selection against Gp-9BB queens parallels the previously demonstrated association between Pgm-3 genotype and the ability of queens to become egg-layers. Because Pgm-3 and Gp-9 are tightly linked and exhibit strong gametic phase disequilibrium ( Ross, 1997), differential acceptance of queens of alternate Pgm-3 and Gp-9 genotypes in polygyne colonies in the introduced range may not be the result of selection acting independently on these two loci.

The objective of the present study was to disentangle the relative roles of Pgm-3 and Gp-9 and/or other closely linked genes on the phenotype and reproductive success of queens produced in polygyne nests of introduced S. invicta. This was accomplished by examining the relationship between the genotypes at the two loci and (i) the weight of mature winged queens, (ii) the rate of wing shedding (dealation) and (iii) the acceptability of young queens initiating reproduction by their mother colony. All experiments were conducted with individuals collected from a Georgia (USA) S. invicta population that has been the subject of an extensive series of population genetics studies. We also made use of the recent discovery that females occasionally are triploid in this population ( Kriegeret al., in press) to dissect in finer detail the effects of Pgm-3 and Gp-9 on queen phenotype and reproductive success.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

One-hundred large polygyne colonies, collected in May and June of 1996 and 1997 from Walton County, Georgia, USA, were established immediately in the laboratory ( Keller & Ross, 1993b). Thirty-nine colonies containing a minimum of four egg-laying queens and large numbers of queen pupae and winged queens were selected for study. The number of egg-laying queens in each of these colonies was reduced to four ( Keller & Ross, 1993a), and all sexuals except for 40 adult winged queens with unsclerotized cuticles (0–3 days old) were removed from each colony. Four days later, all winged queens were removed from each of 19 of these colonies, and 18–20 such queens were placed individually in small fragments of the source colony containing ≈300 worker brood and adults. After 3 days of separation, the queens (then 7–10 days old) were weighed, and we recorded which of them had shed at least one wing (dealated), a behaviour linked to the onset of reproduction ( Fletcher & Blum, 1981; Vargo & Laurel, 1994). All of the queens then were returned individually to their parent colony, and we observed whether or not they were attacked by workers. A previous study showed that worker attacks that are not interrupted invariably lead to the death of a queen within 15 min ( Keller & Ross, 1998). Workers’ responses to introduced queens were determined without knowledge of queen Gp-9 or Pgm-3 genotypes, which were determined later by means of starch gel electrophoresis ( Shoemaker et al., 1992 ). The same procedures were followed for the remaining 20 colonies, except that winged queens were separated from their mother colony after 8 rather than 4 days. Thus, these queens were 11–14 days old when reintroduced into their parent colony.

Ploidy levels were estimated from Gp-9 and Pgm-3 (which are monomeric protein-coding loci) by identifying individuals with two-banded staining patterns that showed uneven band intensities ( Dybdahl & Lively, 1995; Kriegeret al., in press). Gp-9 is a particularly useful marker for detecting triploidy in polygyne S. invicta because of the very high frequency of females bearing both alleles at this locus ( Ross, 1997; Kriegeret al., in press).

Weights of queens were compared using two- and three-way ANOVAs. We considered the effects of the colony of origin in these analyses because previous studies showed significant variation among colonies in the weights of winged queens produced ( Keller & Ross, 1993a, b, 1995). In none of the analyses was the interactions between the colony of origin and either Pgm-3 or Gp-9 genotype significant. We therefore present the results of the ANOVAs without the interaction terms. When there was a significant effect of genotype on weight, we compared pairs of means using both the Scheffé method ( Sokal & Rohlf, 1995) and two-way ANOVAs (which controlled for colony effects). All means are given ± SD.

The significance of associations between genotype and frequency of dealation, and between genotype and proportion of individuals attacked, were determined with G-tests ( Sokal & Rohlf, 1995) or, when cell counts were very low, Fisher’s exact tests.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Weight

7–10-day-old queens

Queens of different Gp-9 and different Pgm-3 genotypes exhibited marked differences in weight (Fig. 1). A two-way ANOVA showed that both Gp-9 (F = 219.37; d.f. = 2; P < 0.001) and Pgm-3 (F = 63.34; d.f. = 2; P < 0.001) genotypes were associated with weight differences when these two loci were considered separately. However, when the effects of both loci were considered simultaneously in a three-way ANOVA, the weight of queens remained highly significantly associated with Gp-9 genotype (F = 81.77; d.f. = 2; P < 0.001) but not with Pgm-3 genotype (F = 0.13; d.f. = 2; NS). Consistent with the view that Pgm-3 has no effect on the weight of queens, the interaction between Gp-9 and Pgm-3 was not significant (F = 0.04; d.f. = 1; NS). In all these analyses, the colony of origin had a significant effect on weight (Gp-9: F = 7.41, d.f. = 18, P < 0.001; Pgm-3: F = 5.85, d.f. = 18, P < 0.001; both loci: F = 7.32, d.f. = 18, P < 0.001).

image

Figure 1 Mean fresh weight (±SD) of 7–1. 0-day-old polygyne S. invicta queens of different Gp-9 and different Pgm-3 genotypes.

Download figure to PowerPoint

That Pgm-3 has no effect on weight once the effect of Gp-9 is taken into account is further supported by comparing the weights of queens having the same Gp-9 genotype but alternative Pgm-3 genotypes. All Gp-9BB queens were of similar weight, regardless of their Pgm-3 genotype (Pgm-3AA = 13.5 ± 1.2 mg, n = 50; Pgm-3Aa = 13.7 ± 1.7 mg, n = 31; Pgm-3aa=13.5 ± 0.9 mg, n = 9, F = 0.11, d.f. = 2, NS). Similarly, Gp-9Bb queens with genotype Pgm-3Aa were very similar in weight to those with genotype Pgm-3aa (11.0 ± 1.1 mg, n = 205 and 11.1 ± 1.1 mg, n = 70, respectively; F = 0.01, d.f. = 1, NS). (Note that the two-locus genotypes Gp-9Bb/Pgm-3AA, Gp-9bb/Pgm-3AA and Gp-9bb/Pgm-3Aa are lacking because gametes with haplotype Gp-9b/Pgm-3A apparently are completely absent in introduced populations ( Ross, 1997).)

Schefféa posteriori tests showed that queens of each of the three Gp-9 genotype classes differed significantly in weight from queens of the other two classes (all P < 0.001) (Fig. 1). Two-way ANOVAs controlling for colony effects gave the same results, with significant differences in queen weight between each pair of Gp-9 genotypes (Gp-9BB vs. Gp-9Bb queens, F = 322.5, d.f. = 1, P < 0.001; Gp-9BB vs. Gp-9bb, F = 49.9, d.f. = 1, P < 0.001; Gp-9Bb vs. Gp-9bb, F = 44.8, d.f. = 1, P < 0.001).

The mean weight of Gp-9Bb queens was approximatively intermediate to those of the two Gp-9 homozygotes (Fig. 1). To quantify the degree of dominance of the Gp-9B allele with respect to queen weight, we used the Dominance Index (D) developed by Wright ( Wright, 1934; Kacser & Burns, 1981). Using the mean weight of polygyne winged queens of each of the three genotypes (WGp-9BB, WGp-9Bb and WGp-9bb) as a measure of the phenotypic values, D is calculated as:

inline image

This index varies between 0 and 1, with a value of 0 indicating that allele Gp-9B is completely dominant, 0.5 indicating that the two alleles are perfectly codominant and 1.0 indicating that Gp-9B is completely recessive. Using our weight data for 7–10-day-old queens yields a value of D = 0.49, indicating nearly perfect codominance of the two Gp-9 alleles with respect to their effect on queen weight.

11–14-day-old queens

There were only 11 Gp-9BB and three Pgm-3AA queens among the 348 queens of this age class, presumably because workers had already eliminated a substantial number of such queens as they approached sexual maturity ( Keller & Ross, 1998; see Discussion). Because of the very low number of Pgm-3AA queens and the complete lack of an effect of Pgm-3 genotype on weight for 7–10-day-old queens, we considered only locus Gp-9 in the following analyses.

Twenty-nine of the 348 queens (7.7%) exhibited uneven band intensities at Gp-9 and/or Pgm-3, indicating that they were triploid. Weight of queens was significantly associated with Gp-9 genotype both when triploids were included in the analysis (F = 55.15, d.f. = 4, P < 0.001) and when they were excluded (F = 72.50, d.f. = 2, P < 0.001) (Fig. 2). As in previous analyses, weight was influenced also by colony of origin (F = 5.67, d.f. = 19, P < 0.001 and F = 5.83, d.f. = 19, P < 0.001 with triploids included and excluded, respectively).

image

Figure 2.   Mean fresh weight (±SD) of 11–14-day-old polygyne S. invicta queens of different Gp-9 genotypes. Individuals in the two columns furthest to the right are confirmed triploids.

Download figure to PowerPoint

Scheffé tests showed that diploid and triploid queens of a given genotype typically differed significantly in weight from queens of all other genotypes (P < 0.001), the exceptions being the comparisons between Gp-9BB and Gp-9BBb queens and between Gp-9Bb and Gp-9Bbb queens (both P > 0.05). The absence of weight differences between queens with these two pairs of genotypes coupled with significant differences between all other pairs was confirmed by two-way ANOVAs controlling for colony effects (P < 0.001 except for Gp-9BB vs. Gp-9BBb and Gp-9Bb vs. Gp-9Bbb queens, both P > 0.05).

The estimated degree of dominance of Gp-9B for presumed diploid 11–14-day-old queens was D = 0.60. Given the relatively high sampling error due to the small number of Gp-9BB queens in this age class, this result is broadly consistent with the conclusion from the 7–10-day-old queens that the two Gp-9 alleles are nearly codominant with regard to their effect on queen weight.

Dealation (wing shedding)

7–10-day-old queens

Sixty-two of the 370 queens (16.8%) dealated. There was a strong association between both Gp-9 (G = 185.5, d.f. = 2, P < 0.001) and Pgm-3 (G = 18.4, d.f. = 2, P < 0.001) genotypes and the probability of dealation. However, inspection of Table 1 and the statistical analyses reveal that all effects can be explained by Gp-9 genotype alone: there was no significant effect of Pgm-3 genotype on rate of dealation among queens with genotype Gp-9BB (G = 0.5, d.f. = 2, NS) nor among queens with genotype Gp-9Bb (G = 2.3, d.f. = 1, NS). In contrast, there was a significant effect of Gp-9 genotype on dealation both among Pgm-3Aa queens (G = 88.7, d.f. = 1, P < 0.001) and among Pgm-3aa queens (G = 198.2, d.f. = 2, P < 0.001).

Table 1.    Proportions of young queens dealating (shedding their wings) according to age and to genotypes at Gp-9 and Pgm-3. Numbers in parentheses indicate sample sizes. Dashes indicate two-locus genotypes that do not occur in the wild ( Ross, 1997) or were not obtained in this study. Thumbnail image of

The strong effect of Gp-9 genotype was due primarily to a high rate of dealation by Gp-9BB queens (66%) compared to queens of the other two genotypes (Gp-9Bb: 1%; Gp-9bb: 0%) (Table 1). The similarly low frequencies of dealation by Gp-9Bb and Gp-9bb queens did not differ significantly (exact test, NS).

11–14-day-old queens

Twenty-two of the 346 queens (6.4%) dealated. As was true for 7–10-day-old queens, there was a strong association between both Gp-9 (G = 44.3, d.f. = 2, P < 0.001) and Pgm-3 (G = 17.1, d.f. = 2, P < 0.001) genotypes and the rate of dealation (Table 1). However, there again was no effect of Pgm-3 genotype once the effect of Gp-9 genotype was considered. Rate of dealation was not significantly associated with Pgm-3 genotype for Gp-9BB queens (G = 1.4, d.f. = 1, NS) nor for Gp-9Bb queens (G = 0.1, d.f. = 1, NS). In contrast, there was a significant effect of Gp-9 genotype among Pgm-3Aa queens (G = 29.7, d.f. = 1, P < 0.001). The effect of Gp-9 genotype on dealation of Pgm-3aa queens was not significant (G = 1.5, d.f. = 1, NS), but there were no Gp-9BB queens in this sample.

As was also true for the younger queens, the rates of dealation for 11–14-day-old queens with the genotypes Gp-9Bb and Gp-9bb (4% and 0%, respectively) were very much lower than for Gp-9BB queens of the same age (82%) (Table 1). Again, dealation rates did not differ significantly between queens of the former two genotypes (exact test, NS).

Because of the lack of an effect of Pgm-3 genotype on dealation in diploid queens, in triploid individuals we tested the association of rate of dealation with Gp-9 genotype only. Four out of the 15 Gp-9BBb queens dealated (27%), whereas not a single Gp-9Bbb queen did so (n = 24), a difference that is highly significant (exact test, P = 0.015). Comparing triploids to diploids, the proportion of Gp-9BBb queens that dealated was significantly lower than the proportion of Gp-9BB queens (exact test, P = 0.013), was not significantly different from the proportion of Gp-9bb queens (exact test, NS) and was significantly greater than the proportion of Gp-9Bb queens that did so (exact test, P = 0.005). The proportion of Gp-9Bbb queens that dealated was significantly lower than the proportion of Gp-9BB queens (exact test, P < 0.001), was not significantly different from the proportion of Gp-9Bb queens (exact test, NS) and was identical to the proportion of Gp-9bb queens that did so.

A comparison between 7–10- and 11–14-day-old diploid queens suggests a general increase in the rate of dealation with age. The older Gp-9BB queens dealated ≈20% more frequently than the younger queens with the same genotype (Table 1). The difference was not significant (exact test, NS), but this may be due to the low number of 11–14-day-old Gp-9BB queens (n = 11). Similarly, the older Gp-9Bb queens were four times more likely to dealate than the younger queens with the same genotype, the difference in this case being significant (G = 5.3, d.f. = 1, P = 0.02). There is no evidence of an effect of age on the rate of dealation for queens of the third genotype (Gp-9bb), as none of these queens shed their wings.

Aggression

7–10-day-old queens

Fifty-five of the 370 queens were attacked by workers when reintroduced into their parental colony (Table 2). There was a strong association between both Gp-9 (G = 190.8, d.f. = 2, P < 0.001) and Pgm-3 (G = 61.7, d.f. = 2, P < 0.001) genotypes and the probability that a queen would be attacked. However, as with weight and dealation rate, the proportion of queens attacked was independent of Pgm-3 genotype once the effect of Gp-9 was taken into account. Gp-9BB queens with alternate Pgm-3 genotypes did not differ significantly in their probability of being attacked (G = 1.1, d.f. = 2, NS); similarly, rates of attacks against Gp-9Bb queens did not vary with their Pgm-3 genotype, as none of these queens was attacked. In contrast, there was a significant effect of Gp-9 genotype on attacks against both Pgm-3Aa queens (G = 90.8, d.f. = 2, P < 0.001) and Pgm-3aa queens (G = 25.5, d.f. = 2, P < 0.001). In fact, all attacked queens possessed genotype Gp-9BB.

Table 2.    Proportions of young queens that were attacked by workers according to queen age and to genotypes at Gp-9 and Pgm-3. Numbers in parentheses indicate sample sizes. Dashes indicate two-locus genotypes that do not occur in the wild or were not obtained in this study. Thumbnail image of
11–14-day-old queens

Ten out of the 346 queens were attacked by workers (Table 2) and, as with 7–10-day-old queens, all of these queens possessed genotype Gp-9BB, leading to a strong association between the proportion of queens attacked and their Gp-9 genotype (G = 89.8, d.f. = 2, P < 0.001). Again, Pgm-3 genotype was not associated with workers’ reponses toward queens once the effect of Gp-9 genotype was taken into account. Gp-9BB queens with alternative Pgm-3 genotypes did not differ significantly in their probability of being attacked (G = 0.7, d.f. = 1, NS), nor was the rate of attacks against Gp-9Bb queens associated with Pgm-3 genotype, as no Gp-9Bb queens were attacked. In contrast, there was a significant association between Gp-9 genotype and occurrence of attacks among Pgm-3Aa queens (G = 57.8, d.f. = 1, P < 0.001). Although there was no significant association between Gp-9 genotype and frequency of attacks among Pgm-3aa queens (none was attacked), again it is important to note that none of these queens possessed genotype Gp-9BB.

None of the 39 identified triploid queens was attacked by workers, resulting in significantly lower proportions of both Gp-9BBb and Gp-9Bbb queens than Gp-9BB queens that were attacked (G = 23.1, d.f. = 1, P < 0.001 and G = 29.5, d.f. = 1, P < 0.001, respectively). Thus, triploid queens bearing at least one copy of allele Gp-9b resembled diploid Gp-9Bb and Gp-9bb queens with respect to being immune to attack by polygyne workers.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

This study shows that Gp-9 or a closely linked gene(s) has major effects on queen phenotype and queen reproductive success in introduced populations of the fire ant S. invicta. Among the mature queens newly produced in nests of the polygyne form (in which multiple egg-laying queens occur in a nest), queens with genotype Gp-9BB weigh about 65% more than queens with genotype Gp-9bb. Heterozygous individuals are of intermediate weight, and the estimated value of the Dominance Index (D) is close to 0.5, indicating nearly complete codominance of the two Gp-9 alleles with respect to their effect on queen weight. Interestingly, the weight of triploid Gp-9BBb queens is very similar to that of Gp-9BB queens, and the weight of triploid Gp-9Bbb queens is similar to that of Gp-9Bb queens. This suggests that there is a dosage effect on queen weight that is related to the number of copies of the Gp-9B allele that an individual possesses. The very low weight of Gp-9bb queens may be linked to the suspected inviability of these queens in the wild. Perhaps they do not have sufficient energy reserves to survive for long in the field, where conditions are harsher than in the laboratory (see also Ross, 1997), or their low weight may reflect more general developmental problems that render them inviable shortly after adult emergence.

Queens with alternate Gp-9 genotypes also varied greatly in the likelihood that they would dealate (shed their wings), a behaviour associated with the onset of reproduction. The probability of Gp-9BB queens dealating was 20–60 times higher than for Gp-9Bb queens. Although not a single Gp-9bb queen dealated in our experiments, the rate of dealation of these queens was not found to be significantly less than that for Gp-9Bb queens, presumably because of the small sample size of queens with the former genotype. The proportion of triploid Gp-9BBb queens dealating was intermediate between the proportions found for Gp-9BB and Gp-9Bb queens. Although these values should be considered with some caution (owing to the small sample size), this result suggests that the rate of dealation may depend not only on the number of copies of the Gp-9B allele a queen possesses, but also on whether or not she carries copies of the Gp-9b allele.

The phenotypic differences associated with the different Gp-9 genotypes have major fitness consequences for queens in polygyne colonies. Most queens homozygous for allele Gp-9B were attacked by workers, and thus would not become new egg layers in these polygyne societies. In contrast, none of the diploid queens with the alternate genotypes was attacked, and thus these queens would have a good chance of being accepted as new egg layers. Interestingly, triploid queens with the genotype Gp-9BBb also were not attacked, suggesting that the presence of one copy of the Gp-9b allele in queens is necessary and sufficient to prevent aggressive behaviour toward them by workers. Thus, in contrast to its effects on queen weight, allele Gp-9b seems to exhibit complete dominance over allele Gp-9B in terms of its effects on queen acceptance in polygyne nests. Keller & Ross (1998) showed that workers immediately attack nestmate workers that have been rubbed against the cuticle of Gp-9BB queens but not those rubbed against Gp-9Bb queens. This indicates that workers discriminate among queens of alternate genotypes on the basis of transferable odours or contact chemicals present on the cuticle of queens. Moreover, workers discriminate increasingly against Gp-9BB queens as they attain sexual maturity, suggesting that recognition and selective elimination of Gp-9BB queens is triggered by two chemical cues, one signalling a queen’s maturity and the other her Gp-9 genotype. Tight coupling between a queen’s reproductive state and pheromone production has been demonstrated previously in S. invicta ( Fletcher & Blum, 1981), as has the ability of workers to assess the level of pheromone production by individual queens ( Willer & Fletcher, 1986). This body of results suggests that polygyne workers can identify the reproductive state of queens by means of pheromones, and that they become aggressive toward queens approaching sexual maturity if these queens do not carry at least one copy of the Gp-9b allele. Under this interpretation, allele Gp-9b again is seen to be completely dominant with respect to the components of queen pheromonal phenotype that elicit worker tolerance.

Our analyses show that once the effect of Gp-9 genotype is taken into account, Pgm-3 genotype is no longer significantly associated with differences in queen phenotype or the probability of queens being accepted in polygyne colonies. This suggests that the associations of Pgm-3 genotype with weight, dealation rate and probability of acceptance by polygyne colonies previously reported in studies that did not control for the effects of Gp-9 genotype (e.g. Keller & Ross, 1993a, 1995) are due to the strong linkage disequilibrium between Pgm-3 and Gp-9, or to linkage disequilibrium between these and other genes affecting queen phenotype and fitness.

The lack of a direct effect of Pgm-3 genotype on queen phenotype, together with recent findings that queen and worker genotypes at Gp-9 (or linked genes) strongly influence the form of social organization, has important implications concerning the reported lack of association between Pgm-3 genotype and queen phenotype in the monogyne social form (in which a single egg-laying queen occurs in a colony) ( Keller & Ross, 1993a, 1995). Ross & Keller (1998) show that worker preferences both for the number of egg-laying queens per colony and for the Gp-9 genotype of such queens are controlled by worker Gp-9 genotype. Colonies containing any Gp-9Bb workers accept multiple queens, but only if these queens also have genotype Gp-9Bb. In contrast, colonies consisting solely of workers with genotype Gp-9BB accept only a single queen that must also bear genotype Gp-9BB. Thus, all females in monogyne colonies are Gp-9BB homozygotes, while they may possess any of the Pgm-3 genotypes. The fact that Pgm-3 variability in this form exists on the background of uniformity in Gp-9 genotype explains why there is no link between Pgm-3 genotype and monogyne queen phenotype (all monogyne queens are large Gp-9BB queens). Therefore, there is no need to invoke an interaction between genotype and social environment to explain the earlier findings based on Pgm-3 (cf. Keller & Ross, 1993a, 1995).

A final point concerns the issue of whether Gp-9 is directly responsible for the phenotypic differences associated with different queen genotypes as well as the specific responses of workers toward queens of different genotypes or, alternatively, whether these effects are due to one or more genes in strong linkage disequilibrium with Gp-9. In the Georgia polygyne population studied here, all of the effects that we describe can be attributed to the two Gp-9 alleles detected electrophoretically. There is a perfect association between a queen’s genotype and her acceptability to polygyne workers, in that Gp-9BB queens are never accepted by such workers in the wild but instead are killed. This association is demonstrated by the complete absence of egg-laying queens of this genotype among 2562 queens collected from polygyne colonies in the field ( Ross, 1997; Goodisman et al., unpublished results). In our experiments, the proportions of Gp-9BB queens killed were only 61% and 91%, respectively, for 7–10- and 11–14-day-old queens. This can be explained by supposing that workers kill Gp-9BB queens only as these queens attain sexual maturity and that not all queens in our study were fully mature. Thus, workers killed a larger proportion of 11–14- than 7–10-day-old Gp-9BB queens, and the initial proportion of Gp-9BB queens was significantly lower in the 11–14- than in the 7–10-day-old class (G = 4.58; d.f. = 1; P=0.03), as expected if workers had already started to execute Gp-9BB queens as they reached sexual maturity. The view that Gp-9BB queens are gradually eliminated as they mature is further supported by the finding that the proportion of Gp-9BB queens invariably decreases over time in laboratory colonies containing even-age cohorts of adult queens (e.g. Keller & Ross, 1998). Finally, the conclusion that all Gp-9BB queens that do not disperse from their nest ultimately are destroyed by workers when they become fully sexually mature is supported by experiments showing that Gp-9BB queens of the polygyne form collected immediately after a mating flight invariably are killed when introduced into foreign polygyne colonies ( Ross & Keller, 1998).

Although data from Georgia can be explained by assuming that Gp-9 is the locus directly responsible for the effects we report, other lines of evidence militate in favour of additional genes (or additional alleles at Gp-9) being involved. First, polygyne nests in native populations of S. invicta in Argentina may contain Gp-9BB or Gp-9Bb queens, although they never contain any Pgm-3AA queens (some Gp-9BB queens do not possess the Pgm-3AA genotype; Ross, 1997). This pattern previously was explained by suggesting that Pgm-3 is the locus responsible for the differential acceptance of queens in polygyne colonies ( Ross et al., 1996a ; Ross, 1997). However, our experiments show that Pgm-3 genotype does not influence queen acceptance in polygyne colonies in the introduced range once the effect of Gp-9 genotype is taken into account. If Gp-9 is the actual gene involved in the queen phenotypic differences used as cues by executing polygyne workers, this pattern can be explained only if there is an additional allele, say Gp-9b*, in South America that causes the same phenotypic effects as Gp-9b but is indistinguishible from Gp-9B using our standard electrophoretic techniques. This allele presumably would have been lost in the USA populations as a result of the founder event associated with initial colonization ( Ross et al., 1993 ). Under this scenario, native polygyne colonies may contain either Gp-9Bb* or Gp-9Bb reproductive queens (assuming genotype Gp-9b*b* is lethal, as is Gp-9bb). Allele Gp-9b* most likely would be a mutant allele that evolved from allele Gp-9b on a chromosome that also contained allele Pgm-3a. The mutant haplotype Gp-9b*/Pgm-3a thus would give rise to viable Gp-9Bb*/Pgm-3Aa queens with an electrophoretic phenotype indistinguishable from Gp-9BB/Pgm-3Aa, and Gp-9Bb*/Pgm-3aa queens indistinguishable from Gp-9BB/Pgm-3aa.

Alternatively, a gene other than Gp-9 might be responsible for the differential acceptance of queens in polygyne colonies. Under this scenario, an allele ACC+ at this separate gene that confers acceptability to queens in polygyne colonies would be completely associated with Gp-9b in the USA but not in South America, where it would also occur in association with allele Gp-9B (as evidenced by the acceptability of some Gp-9BB polygyne queens). Such differences in gametic disequilibrium between the native and introduced ants again could stem from the initial founder event, such that Gp-9b/ACC+ but not Gp-9B/ACC+ chromosomes were present in the founding population. The observed link between Pgm-3 genotype and queen acceptance could be accounted for if ACC+ is never associated with Pgm-3A (thus explaining the complete lack of Pgm-3AA queens in the USA and South America) and is associated with Pgm-3a in only some proportion of chromosomes (thus accounting for the execution of some Pgm-3Aa and Pgm-3aa queens in the USA; Ross & Keller, 1998).

Another line of evidence favouring the involvement of genes other than Gp-9 in producing the phenotypic effects reported here is that Gp-9 and Pgm-3 may belong to a large genomic region with little or no recombination, such as an inversion (e.g. Ross et al. 1996a ). Gp-9 and Pgm-3 are very tightly linked ( Ross, 1997), even though S. invicta possesses a haploid chromosome complement of n = 16 ( Glancey et al., 1976 ) and fewer than 35 polymorphic loci have been surveyed in this species. The probability of finding two tightly linked loci under such circumstances in the absence of an inversion or other suppressor of recombination must be very low. Moreover, segregation distorters, which have evolutionary dynamics similar in many respects to Gp-9 ( Keller & Ross, 1998), often are associated with inversions ( Hammer et al., 1989 ; Lyttle, 1993; Silver, 1993). The presence of a region including Gp-9 and Pgm-3 that does not undergo frequent recombination would mean that potentially many genes of substantial effect could be responsible for the phenotypic and fitness differences ascribed to variation at these two genes.

Although differences in Gp-9 genotype distributions in South America and the USA point to the involvement of supplemental alleles or loci, it is also possible that Gp-9B and Gp-9b do in fact underly the effects we report, but that selection pressures relating to worker acceptance of queens in polygyne nests differ somewhat between the two areas. Under this scenario, some ecological, genetic or social factors present in the native but not the introduced range may favour the occasional retention of Gp-9BB queens by polygyne workers.

In conclusion, current data suggest that a genetic element inherited as a single Mendelian factor controls a large suite of traits important in the social organization of S. invicta colonies. This element is more closely associated with the marker gene Gp-9 than the marker gene Pgm-3 and, in the introduced range, appears to be in complete linkage disequilibrium with Gp-9. In the native populations this is not the case. Feasible explanations of the data from both the native and the introduced ranges invoke the presence of cryptic variation at Gp-9 or the presence of additional genes of major effect in disequilibrium with Gp-9, coupled with the loss of allelic or haplotypic variation during initial colonization of the USA. Alternatively, distinct selection pressures relating to colony genotypic composition may exist in the two ranges. Future phenotypic and genetic studies of South American and USA samples will address this important issue of the specific genetic architecture underlying social divergence in fire ants.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We thank Chris DeHeer, Mike Goodisman, Mark Mescher and one reviewer for comments on the manuscript. This work was funded by grants from the Swiss and US National Science Foundations and the National Geographic Society, and by the Georgia Agricultural Experiment Stations, University of Georgia.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • 1
    Barton, N.H. & Turelli, M. 1989. Evolutionary quantitative genetics: how little do we know? Ann. Rev. Genet. 23: 337 370.
  • 2
    Boake, C.R.B., ed. 1994. Quantitative Genetic Studies of Behavioral Evolution. University of Chicago Press, Chicago.
  • 3
    Charlesworth, B. 1990. The evolutionary genetics of adaptation. In: Evolutionary Innovations (M. H. Nitecki, ed.), pp. 47–70. University of Chicago Press, Chicago.
  • 4
    Clark, A.G. & Koehn, R.K. 1991. Enzymes and adaptation. In: Genes in Ecology (R. J. Berry, T. J. Crawford & G. M. Hewitt, eds), pp. 193–227. Blackwell Scientific, Oxford.
  • 5
    Dybdahl, M.F. & Lively, C.M. 1995. Diverse, endemic and polyphyletic clones in mixed populations of a freshater snail (Potamopyrgus antipodarum) . J. Evol. Biol. 8: 385 398.
  • 6
    Feder, M.E. & Watt, W.B. 1991. Functional biology of adaptation. In: Genes in Ecology (R. J. Berry, T. J. Crawford & G. M. Hewitt, eds), pp. 365–392. Blackwell Scientific, Oxford.
  • 7
    Ferguson, M.M. & Danzmann, R.G. 1985. Pleiotropic effects of a regulatory gene (Pgm-1-t) on the social behavior of juvenile rainbow trout (Salmo gairdneri) . Can. J. Zool. 63: 2847 2851.
  • 8
    Fletcher, D.J.C. & Blum, M.S. 1981. Pheromonal control of dealation and oogenesis in virgin queen fire ants. Science 212: 73 75.
  • 9
    Glancey, B.M., St Romain, M.K., Crozier, R.H. 1976. Chromosome numbers of the red and black fire ants, Solenopsis invicta and S. richteri. Ann. Ent. Soc. Am. 69: 469 470.
  • 10
    Hammer, M.F., Schimenti J., Silver, L.M. 1989. Evolution of mouse chromosome 17 and the origin of inversions associated with t haplotypes. Proc. Natl. Acad. Sci. USA 86: 3261 3265.
  • 11
    Hoffmann, A.A. 1994. Behaviour genetics and evolution. In: Behaviour and Evolution (P. J. B. Slater, & T. R. Halliday, eds), pp. 7–42. Cambridge University Press., Cambridge.
  • 12
    Hunt, G.J., Guzman Novoa, E., Fondrk M.K., Page, R.E. 1998. Quantitative trait loci for honey bee stinging behavior and body size. Genetics 148: 1203 1213.
  • 13
    Hunt, G.J., Page, R.E., Fondrk, M.K., Dullum, C.J. 1995. Major quantitative trait loci affecting honey bee foraging behavior. Genetics 141: 1537 1545.
  • 14
    Kacser, H. & Burns, J.A. 1981. The molecular basis of dominance. Genetics 97: 639 666.
  • 15
    Keller, L. & Ross, K.G. 1993a. Phenotypic basis of reproductive success in a social insect: genetic and social determinants. Science 260: 1107 1110.
  • 16
    Keller, L. & Ross, K.G. 1993b. Phenotypic plasticity and cultural transmission of alternative reproductive strategies in the fire ant Solenopsis invicta. Behav. Ecol. Sociobiol. 33: 121 129.
  • 17
    Keller, L. & Ross, K.G. 1995. Gene by environment interaction – effects of a single gene and social environment on reproductive phenotypes of fire ant queens. Funct. Ecol. 9: 667 676.
  • 18
    Keller, L. & Ross, K.G. 1998. Selfish genes: a green beard in the red fire ant. Nature 394: 573 575.
  • 19
    Krieger, M.J.B., Ross, K.G., Chang C.W., Keller, L. in press. Frequency and origin of triploidy in the fire ant Solenopsis invicta. Heredity.
  • 20
    Lank, D.B., Smith, C.M., Hanotte, O., Burke T., Cooke, F. 1995. Genetic polymorphism for alternative mating behaviour in lekking male ruff, Philomachus pugnax. Nature 378: 59 62.
  • 21
    Lyttle, T.W. 1993. Cheaters sometimes prosper: distortion of mendelian segregation by meiotic drive. Trends in Genetics 9: 205 210.
  • 22
    Mitchell-Olds, T. 1995. The molecular basis of quantitative genetic variation in natural populations. Trends Ecol. Evol. 10: 324 328.
  • 23
    Moritz, R.F.A. & Hillesheim, E. 1985. Inheritence of dominance in honeybees (Apis mellifera capensis Esch.). Behav. Ecol. Sociobiol. 17: 87 89.
  • 24
    Page, R. & Robinson, G. 1991. The genetics of division of labour in honey bee colonies. Adv. Ins. Physiol. 23: 118 169.
  • 25
    Page, R. & Robinson, G. 1994. Reproductive competition in queenless honey bee colonies (Apis mellifera) . Behav. Ecol. Sociobiol. 35: 99 107.
  • 26
    Robinson, G.E. in press. Integrative animal behavior and sociogenomics. Trends Ecol. Evol.
  • 27
    Robinson, G.E., Fahrbach S.E., Winston, M.L. 1997. Insect societies and the molecular biology of social behavior. Bioessays 19: 1099 1108.
  • 28
    Robinson, G.E. & Page, R.E. 1995. Genotypic constraints on plasticity for corpse removal in honey bee colonies. Anim. Behav. 49: 867 876.
  • 29
    Ross, K.G. 1992. Strong selection on a gene that influences reproductive competition in a social insect. Nature 355: 347 349.
  • 30
    Ross, K.G. 1997. Multilocus evolution in fire ants – effects of selection, gene flow and recombination. Genetics 145: 961 974.
  • 31
    Ross, K.G. & Keller, L. 1995. Ecology and evolution of social organization – insights from fire ants and other highly eusocial insects. Ann. Rev. Ecol. Syst. 26: 631 656.
  • 32
    Ross, K.G. & Keller, L. 1998. Genetic control of social organization in an ant. Proc. Natn. Acad. Sci. USA 95: 14232 14237.
  • 33
    Ross, K.G., Vargo E.L., Keller, L. 1996a. Simple genetic basis for important social traits in the fire ant Solenopsis invicta. Evolution 50: 2387 2399.
  • 34
    Ross, K.G., Vargo E.L., Keller, L. 1996b. Social evolution in a new environment: The case of introduced fire ants. Proceedings Natn. Acad. Sci., USA 93: 3021 3025.
  • 35
    Ross, K., Vargo, E., Keller L., Trager, J. 1993. Effect of a founder event on variation in the genetic sex-determining system of the fire ant Solenopsis invicta. Genetics 135: 843 854.
  • 36
    Shaw, K.L. 1996. Polygenic inheritance of a behavioral phenotype: interspecific genetics of song in the Hawaiian cricket genus Laupala. Evolution 50: 256 266.
  • 37
    Shoemaker, D.D., Costa, J.T., Ross, G.K. 1992. Estimates of heterozygosity in two social insects using a large number of electrophoretic markers. Heredity 69: 573 582.
  • 38
    Shuster, S.M. & Sassaman, C. 1997. Genetic interaction between male mating strategy and sex ratio in a marine isopod. Nature 388: 373 377.
  • 39
    Silver, L.M. 1993. The peculiar journey of a selfish chromosome: mouse t haplotypes and meiotic drive. Trends Genet 9: 250 254.
  • 40
    Sokal, R.R. & Rohlf, F.J. 1995. Biometry. The Principles and Practice of Statistics in Biological Research, 3rd edn. Freeman, San Francisco.
  • 41
    Vargo, E.L. & Laurel, M. 1994. Studies on the mode of action of a queen primer pheromone of the fire ant Solenopsis invicta. J. Insect. Physiol. 40: 601 610.
  • 42
    Willer, D.E. & Fletcher, D.J.C. 1986. Differences in inhibitory capability among queens of the ant Solenopsis invicta. Physiol. Entomol. 11: 475 482.
  • 43
    Wright, S. 1934. Physiological and evolutionary theories of dominance. Am. Nat. 68: 24 53.