Worker Reproduction and Caste Polymorphism Impact Genome Evolution and Social Genes Across the Ants

Abstract

Eusocial insects are characterized by several traits, including reproductive division of labor and caste polymorphisms, which likely modulate genome evolution. Concomitantly, evolution may act on specific genes and pathways underlying these novel, sociality-associated phenotypes. Reproductive division of labor should increase the magnitude of genetic drift and reduce the efficacy of selection by reducing effective population size. Caste polymorphism has been associated with relaxed selection and may facilitate directional selection on caste-specific genes. Here, we use comparative analyses of 22 ant genomes to test how reproductive division of labor and worker polymorphism influence positive selection and selection intensity across the genome. Our results demonstrate that worker reproductive capacity is associated with a reduction in the degree of relaxed selection but is not associated with any significant change to positive selection. We find decreases in positive selection in species with polymorphic workers, but no increase in the degree of relaxed selection. Finally, we explore evolutionary patterns in specific candidate genes associated with our focal traits in eusocial insects. Two oocyte patterning genes previously implicated in worker sterility evolve under intensified selection in species with reproductive workers. Behavioral caste genes generally experience relaxed selection associated with worker polymorphism, whereas vestigial and spalt, both associated with soldier development in Pheidole ants, experience intensified selection in worker polymorphic species. These findings expand our understanding of the genetic mechanisms underlying elaborations of sociality. The impacts of reproductive division of labor and caste polymorphisms on specific genes illuminate those genes' roles in generating complex eusocial phenotypes.

Keywords: comparative genomics; eusocial insect evolution; positive selection; selection intensity.

Fig. 1.

Predicted and recovered effects of worker reproduction and worker polymorphism on genome-wide evolutionary patterns. Tan color and upward arrows indicate an increase in the number of genes that experience a selective regime in species with the focal trait relative to those without; blue color and downward arrows indicate a decrease in the number of genes experiencing that selective regime, with a lighter color and N.S. notation to note nonsignificant results based on a bootstrapping test; and gray color and dash indicate no significant change.

Fig. 2.

Phylogeny of sampled ant species, illustrating character states for focal social traits. Dots at the phylogeny tips are colored to indicate the trait state for that species for worker reproduction and worker polymorphism. Phylogeny adapted from Blanchard and Moreau 2017; sources for character states can be found in supplementary table S1, Supplementary Material online.

Fig. 3.

Impact of sociality-associated traits on patterns of positive selection across ant genomes. For each trait, the count of orthogroups experiencing positive selection in association with trait presence (dark-red bars) and the count of orthogroups experiencing positive selection in association with trait absence (light-red bars) are shown in the top panel. To assess the potential impact of species sampling on the ability to detect positive selection, we implemented a bootstrapping approach; for each trait, these results are shown in the bottom panel. For both worker reproduction and worker polymorphism, there are fewer genes experiencing positive selection associated with trait presence than with trait absence; however, based on the bootstrapping analysis, we are unable to rule out that this pattern is the result of species sampling, as our recovered ratios fall in the middle of the distribution of bootstrapping results.

Fig. 4.

Patterns of selection intensity genome wide. In the top panel, for each trait, the count of orthogroups experiencing relaxed selection (in lighter blue) and the count of orthogroups experiencing intensified selection (in darker blue) in association with trait presence are shown. In species with reproductive workers, left, there are significantly fewer orthogroups experiencing relaxed selection versus intensified selection (chi-squared goodness-of-fit test, P = 1.89e⁻¹³⁰). In species with polymorphic workers, right, there is no significant difference in the number of orthogroups experiencing relaxed versus intensified selection (chi-squared goodness-of-fit test, P = 0.3741). In the bottom panel, for each trait, the distribution of significant k-values inferred by RELAX is shown. k-values measure the degree of selective intensity, with values less than 1 indicating relaxation of selection, and values greater than 1 indicating intensification. Because k-values can range from zero to fifty, the x-axis is log-transformed for visual clarity.