A shift to shorter cuticular hydrocarbons accompanies sexual isolation among Drosophila americana group populations

Abstract

Because sensory signals often evolve rapidly, they could be instrumental in the emergence of reproductive isolation between species. However, pinpointing their specific contribution to isolating barriers, and the mechanisms underlying their divergence, remains challenging. Here, we demonstrate sexual isolation due to divergence in chemical signals between two populations of Drosophila americana (SC and NE) and one population of D. novamexicana, and dissect its underlying phenotypic and genetic mechanisms. Mating trials revealed strong sexual isolation between Drosophila novamexicana males and SC Drosophila americana females, as well as more moderate bi-directional isolation between D. americana populations. Mating behavior data indicate SC D. americana males have the highest courtship efficiency and, unlike males of the other populations, are accepted by females of all species. Quantification of cuticular hydrocarbon (CHC) profiles-chemosensory signals that are used for species recognition and mate finding in Drosophila-shows that the SC D. americana population differs from the other populations primarily on the basis of compound carbon chain-length. Moreover, manipulation of male CHC composition via heterospecific perfuming-specifically perfuming D. novamexicana males with SC D. americana males-abolishes their sexual isolation from these D. americana females. Of a set of candidates, a single gene-elongase CG17821-had patterns of gene expression consistent with a role in CHC differences between species. Sequence comparisons indicate D. novamexicana and our Nebraska (NE) D. americana population share a derived CG17821 truncation mutation that could also contribute to their shared "short" CHC phenotype. Together, these data suggest an evolutionary model for the origin and spread of this allele and its consequences for CHC divergence and sexual isolation in this group.

Keywords: cuticular hydrocarbons; drosophila; reproductive isolation.

Figure 1

Male courtship behavior rates in 1x1 mating assays across all cross types. Points represent individual trials, with boxes showing quartiles and the mean as a solid bar. Males (columns) show significant differences in display rate and tap rate, and marginal differences in lick rate. Rates did not differ based on female identity (rows) or male × female interaction (see results). Bar and whiskers indicate mean and standard error (SE).

Figure 2

Cuticular hydrocarbon composition of unmanipulated males and females of each species stock. (A) Relative log‐scale abundance of compounds for males (upper) and females (lower), for each major compound type (line = mean, box = central quartile, whiskers = S.E.). n = 5 samples for each sex and line. Test statistics for species differences are in Table S3. (B) First two principal components (U‐PC1 and U‐PC2) of composite CHC variation among males (circle) and females (triangle), with percent of total variance explained. Ellipses indicate 90% bivariate normal density for each species‐sex group. Species and sex significantly influenced both U‐PC1 and U‐PC2.

Figure 3

CHC candidate gene expression, genealogical relationships, and sequence variation among D. americana group lines. (A) Gene expression variation showing mean and standard error between focal species lines for candidate genes with the largest differences in either males (upper, circles) or females (lower, triangles). Gene expression is measured in transcripts per million (TPM, n = 3 for each sample). (B) Gene tree topologies for all 23 candidate genes using either western (NE) or eastern (IN) populations for NE D. americana. (C) Schematic of loci (green/blue boxes) in the genomic window surrounding candidates CG17821 and CG18609. Blue boxes indicate a genealogy that groups the NE D. americana population with D. novamexicana; gene topology of shaded blue region displayed below line. Green boxes indicate loci with genealogies matching the expected species tree. (D) 3’ terminal nucleotides and occurrence of inferred ancestral and derived (truncated) allele in CG17821 among lineages with sequence data.