Adaptive evolution of recently duplicated accessory gland protein genes in desert Drosophila

Abstract

The relationship between animal mating system variation and patterns of protein polymorphism and divergence is poorly understood. Drosophila provides an excellent system for addressing this issue, as there is abundant interspecific mating system variation. For example, compared to D. melanogaster subgroup species, repleta group species have higher remating rates, delayed sexual maturity, and several other interesting differences. We previously showed that accessory gland protein genes (Acp's) of Drosophila mojavensis and D. arizonae evolve more rapidly than Acp's in the D. melanogaster subgroup and that adaptive Acp protein evolution is likely more common in D. mojavensis/D. arizonae than in D. melanogaster/D. simulans. These findings are consistent with the idea that greater postcopulatory selection results in more adaptive evolution of seminal fluid proteins in the repleta group flies. Here we report another interesting evolutionary difference between the repleta group and the D. melanogaster subgroup Acp's. Acp gene duplications are present in D. melanogaster, but their high sequence divergence indicates that the fixation rate of duplicated Acp's has been low in this lineage. Here we report that D. mojavensis and D. arizonae genomes contain several very young duplicated Acp's and that these Acp's have experienced very rapid, adaptive protein divergence. We propose that rapid remating of female desert Drosophila generates selection for continuous diversification of the male Acp complement to improve male fertilization potential. Thus, mating system variation may be associated with adaptive protein divergence as well as with duplication of Acp's in Drosophila.

Figure 1.—

Phylogeny of Acp5 duplicate genes. (A) Evolution along each branch is shown as d_N/d_S ratios. (B) Maximum-likelihood analyses of the complete gene tree. There is no significant evidence of branch heterogeneity; however, the complete Acp5 gene tree has d_N/d_S > 1 (P < 0.01). ω, d_N/d_S ratio; 2Δℓ, likelihood-ratio test; NS, not significant.

Figure 2.—

Phylogeny of Acp16 duplicate genes. (A) d_N/d_S values are shown for each branch. (B) Maximum-likelihood analyses of the complete gene tree. There is no significant evidence of branch heterogeneity; however, the complete Acp16 gene tree has d_N/d_S > 1 (P < 0.05). ω, d_N/d_S ratio; 2Δℓ, likelihood-ratio test; NS, not significant.