Evolution of multiple additive loci caused divergence between Drosophila yakuba and D. santomea in wing rowing during male courtship

Abstract

In Drosophila, male flies perform innate, stereotyped courtship behavior. This innate behavior evolves rapidly between fly species, and is likely to have contributed to reproductive isolation and species divergence. We currently understand little about the neurobiological and genetic mechanisms that contributed to the evolution of courtship behavior. Here we describe a novel behavioral difference between the two closely related species D. yakuba and D. santomea: the frequency of wing rowing during courtship. During courtship, D. santomea males repeatedly rotate their wing blades to face forward and then back (rowing), while D. yakuba males rarely row their wings. We found little intraspecific variation in the frequency of wing rowing for both species. We exploited multiplexed shotgun genotyping (MSG) to genotype two backcross populations with a single lane of Illumina sequencing. We performed quantitative trait locus (QTL) mapping using the ancestry information estimated by MSG and found that the species difference in wing rowing mapped to four or five genetically separable regions. We found no evidence that these loci display epistasis. The identified loci all act in the same direction and can account for most of the species difference.

Figure 1. Courtship behavior in wild type D. yakubaand D. santomea.

(A) Males of both species first orient towards the female and tap her with a T1 leg. Then they approach the back or the side of the female and periodically sing by vibrating one extended wing. Song bouts are punctuated by circling to the side and front of the female (circling is sometimes accompanied by shaking of both wings) or by attempts to lick the female's genitalia. If the female is receptive, a lick is followed immediately by copulation. Cartoons were adapted from movie still images. Courtship steps at which D. santomea males are observed to row are marked with an asterisk (*). (B) Distribution of wing rowing in the species of the D. melanogaster species subgroup. The molecular phylogeny was adapted from Prud'homme et al. . (−) and (+) indicate which species row, and relative rowing frequencies (rows/second courtship): D. mauritiana = 0.0322; D. sechellia = 0.0228; D. simulans = 0.0279; D. teissieri = 0.0061; D. yakuba = 0.0010; D. santomea = 0.1033; D. orena = 0.0144. (C) Chronophotograph of a D. santomea male rowing while positioned behind a stationary female. (D) The frequency of rowing in multiple independent D. santomea and D. yakuba isolates. y-axis: wing rows normalized to seconds of courtship in a 15 minute movie. Sample means are marked by the filled circle and lines indicate +/− one standard deviation. Species level differences D. santomea and D. yakuba lines were highly significant (Nested Anova; D.F. = 4,184; F = 266.4; p<2.2e-16), while only D. santomea STO CAGO 1495-5 rowed significantly more than the other D. santomea isolates (Anova; D.F. = 3, 81; F = 18.26; p<1.62e-06) and there were no significant differences between D. yakuba isolates.

Figure 2. Wing rowing in F1 and backcross hybrid males.

y-axis: wing rows normalized to seconds courtship observed in a 15 minute movie. x-axis from left to right: F1 hybrid males from D. santomea females crossed to D. yakuba males; F1 hybrid males from the reciprocal cross; males from an F1 hybrid female backcrossed to a D. santomea male; males from an F1 hybrid female backcrossed to a D. yakuba male. Levels of wing rowing in F1 hybrid males from either cross direction are not significantly different (Student's t-Test; t = 1.3672; D.F. = 58.86; p = 0.1768; two-tailed) indicating there are no contributing loci on the X chromosome. Filled circles represent the mean level of rowing and lines +/− one standard deviation. Dashed line indicates mean wing rowing frequency for the D. santomea parental line from Fig. 1.

Figure 3. QTL analysis of wing rowing in D. santomea and D. yakuba backcross males.

QTL maps of the D. santomea backcross (A) and D. yakuba backcross (B) flies from Fig. 2. LOD is indicated on the y-axis. The x-axis is the physical map based on the D. yakuba genome, SNP markers are represented as black tick marks. Solid lines represent the LOD scores for wing rows normalized to seconds courtship, dashed lines for normalization to seconds the female was stationary during courtship (WWPSSF). Solid and dashed horizontal lines illustrate the corresponding permutation-determined 0.05 (lower) and 0.01 (upper) confidence limits, respectively. The 0.01 confidence limit in the D. yakuba backcross for stationary female normalization equals 5.02 and is not shown in panel B. In (A), a, b′, b, c and d indicate QTL at positions chr2:8,591,051, chr2:15,687,032, chr2:30,610,169, chr2:43,343,589 and chr3:46,460,168, respectively. In (B), a and b indicate QTL at chr2:8,591,087 and chr2:32,778,893.

Figure 4. Effect plots for wing rowing QTL.

(A) Effect plot for the QTL on chromosome 2 at 15,687,032 bp for the D. santomea backcross. (B) Effect plot for the QTL on chromosome 2 at 43,343,589 bp for the D. santomea backcross. (C) Effect plot for the QTL on chromosome 3 at 46,982,725 bp for the D. santomea backcross. (D) Effect plot for the QTL on chromosome 2 at 8,591,087 bp for the D. yakuba backcross. For each plot, the posterior probability of homozygosity as predicted by the HMM for backcross hybrid flies at the marker for the peak LOD score for each QTL is plotted on the x-axis. Heterozygous flies have a value of 0, homozygous flies a value of 1, and flies where the genotype is uncertain due to low marker density fall between zero and one. The number of wing rows normalized to seconds spent courting is plotted on the y axis.