SEX-RATIO DETERMINATION WITHIN COLONIES OF ANTS.

Abstract

Trivers and Hare (1976) propose that in social Hymenoptera a queen and her daughter workers can be in conflict over the relative amount invested in the two sexes. A worker is three times more related to a full sister than to a brother, while the queen is equally related to both sons and daughters. They present a wide array of evidence which suggests that wherever conflict occurs, the workers are likely to win and produce female-biased sexual investment ratios. Several subsequent studies, however, appear to indicate that, within some individual colonies, queens can exert enough influence to produce sex ratios favorable to their interests (Brian, 1 979; Forsyth, 1981; Pamilo and Rosengren, 1983). Herbers's study (1984) is perhaps the most thorough investigation of worker-queen conflict within a single hymenopteran species. She presents two major lines of evidence suggesting that workers of the ant species Leptothorax longispinosus attempt to maximize inclusive fitness at the expense of their queen's and that an active conflict occurs within colonies. First, proportional investment in females by colonies follows the order predicted by kin selection where, queenless > monogynous > oligogynous (2-4 functional queens) > polygynous (?5 queens). Queenless colonies are the most femalebiased, because the workers could produce their optimal sex ratio without opposition from a queen. Oligogynous and polygynous colonies are the least female-biased, because increasing queen number reduces the asymmetry in relatedness between brothers and sisters, and queen and worker investment equilibria converge (Trivers and Hare, 1976). Also, the queen-to-worker ratio is increased, which should favor the queens doing better in conflict situations. The second line of evidence is that worker number is often correlated negatively with proportional investment in males for monogynous, oligogynous, and polygynous colonies. This supports the conflict hypothesis in that the larger the work force, the more difficult it may be for a queen to affect the sex ratio. Overall, Herbers's results and the variance in investment ratios between colonies found by numerous studies (see Nonacs [ 1986] for a review), appear to suggest that a dynamic conflict occurs between workers and queens over sex ratio. Male production seems to be increased in direct proportion to the ability of queens to succeed in the conflicts. I propose, however, that the sex ratios produced by individual colonies primarily reflect the workers' interests. I will support this hypothesis in two ways. First, I will show that Herbers's study (as the particular case in point) and the variance in investment ratios from many other ant species are equally well (or better) explained by a proximate mechanism of resource availability as by worker-queen conflict. Second, I will argue that because the fitness of any given colony's investment ratio depends on the sex ratio of the interbreeding population, colonies producing male-biased broods can still be harmonious with a hypothesis of total worker control although, superficially, they may appear queen-influenced. The patterns of investment observed by Herbers, particularly in reference to worker number, may also be explained by a resource availability hypothesis. Larger colonies with more workers may generally be expected to collect more food (Oster and Wilson, 1978 pp. 26-74; Brian et al., 1981; Elmes and Wardlaw, 1982). It is reasonable to assume that with more food, sexual production increases. In many species, investment in males is negatively correlated with total sexual production (Nonacs, 1986). In other words, with increasing resources, colonies are more likely to favor female production. Therefore, worker number may be only spuriously correlated to investment ratios due to its relationship to ergonomic efficiency. Available data sets from 18 monogynous, five polygynous, and one slave-making species measure sexual production and worker number. These can be tested as to whether one, both, or neither accurately predicts sex ratio. For each species, I have examined the partial correlations among the total amount of sexuals produced (measured as biomass), the number of adult workers, and the male proportion of the total sexual biomass (R), which was normalized by arcsine square root transformation. The relationship between total sexual biomass and proportion invested in males is strong (Table 1). When worker number is controlled, seven correlations are significantly negative (P < 0.025). In total, 20 of 24 correlations are negative, a significant trend across species (sign test, P < 0.002). With total sexual biomass controlled, however, there is almost no relationship between worker number and proportion invested in males (Table 1). In only two species are the correlations significant, with one being negative and the other positive. Moreover, there is no trend across species as the correlations are equally split between negative and positive (12 of 24 negative). When R is controlled, there are positive corre-