Evolution of a neuromuscular sexual dimorphism in the Drosophila montium species group

Abstract

While epigamic traits likely evolve via sexual selection, the mechanism whereby internal sexual dimorphism arises remains less well understood. Seeking clues as to how the internal sexual dimorphism evolved, we compared the abdominal musculature of 41 Drosophila montium group species, to determine whether any of these species carry a male-specific muscle of Lawrence (MOL). Our quantitative analysis revealed that the size of a sexually dimorphic MOL analog found in 19 montium group species varied widely from species to species, suggesting the gradual evolution of this sexually dimorphic neuromuscular trait. We attempted the ancestral state reconstitution for the presence or absence of the neuromuscular sexual dimorphism in the A5 segment; the neuromuscular sexual dimorphism existed in an old ancestor of the montium group, which was lost in some of the most recent common ancestors of derived lineages, and subsequently some species regained it. This loss-and-gain history was not shared by evolutionary changes in the courtship song pattern, even though both traits were commonly regulated by the master regulator male-determinant protein FruM. It is envisaged that different sets of FruM target genes may serve for shaping the song and MOL characteristics, respectively, and, as a consequence, each phenotypic trait underwent a distinct evolutionary path.

Figure 1

Detection of sexual dimorphism in the abdominal musculature of D. melanogaster and some species of the montium group. Representative images (a1–a3,b1,b2,c1,c2,d1,d2) of abdominal muscles and the Gaussian fit to the F_A/F_B distribution of the largest muscle (a4,b3,c3,d3) in the reference species D. melanogaster (a) and 3 species of the montium groups, i.e., leontia (b), lactericornis (c) and cauverii (d). Images are typical examples of male abdominal musculatures with the exception of those for D. melanogaster, which are images of a wild-type male (a1), wild-type female (a2) and fru mutant male (a3). The MOL and its analogs are indicated with arrowheads, and the regions shown in Fig. 4 to visualize nuclei are boxed. D. cauverii is a species that did not show the neuromuscular sexual dimorphism in the A5 segment whereas 3 other species displayed the sexual dimorphism. Scale bar: 100 µm. The curves in a4, b3, c3 and d3 compare F_A/F_B distributions for the male A5 (red lines), the female (black lines) and male A4 (blue lines), and, only in a4, the male fru mutant A5 (magenta lines). The statistical differences were evaluated by the Brown-Forsythe and Welch ANOVA or Kruskal–Wallis test; ***P < 0.001.

Figure 2

Sexually dimorphic size variations of the largest A5 muscles in D. kikkawai. (a–f) MOL-analogs are visually recognizable bilaterally (a–c) or unilaterally (d) in most males but not females (f) and some males (e) from A5. (g) The scatter plot shows the distribution of mean F_A/F_B values as compared between the male and female. Each symbol represents the values estimated for a hemi-segment. The red and blue long lines represent the mean values for female and male values, respectively. The short bar represents the SEM. The blue and pink areas in the graph indicate the 99% confidence intervals (mean ± 3 × SD values) of the muscle size for males and females, respectively.

Figure 3

Divergent neuromuscular sexual dimorphisms in A5 among species of the montium group. F_A/F_B values in the 41 species examined are shown on the phylogenetic tree. The scatter plot shows the distribution of mean F_A/F_B values as compared between the male and female. Each symbol represents the values estimated for a hemi-segment. The red and blue long lines represent the mean values for female and male values, respectively. The short bar represents the SEM. The blue and pink areas in the graph indicate the 99% confidence intervals (mean ± 3 × SD values) of the muscle size for males and females, respectively. The neuromuscular system in the A5 segment is sexually dimorphic (+) or sexually monomorphic (−). Courtship song-types reported in our previous paper (Chen et al.) are indicated in the right-hand side of the plots. The song types for pre‐ and post‐mounting songs are shown separated by a hyphen (pre/post): P1–P3, pulse songs (P) distinct in certain song parameters; HPR, high pulse repetition song; S1 and S2, sine songs (S) distinct in certain song parameters; –, no song.

Figure 4

Inter-species variations in the number of fibers composing an MOL or MOL-analog. White dots represent the position of nuclei recorded on a transparency (see Materials and Methods), which aligns along a longitudinal axis, visualizing a single fiber composing the MOL or MOL-analog. Shown are representative examples of MOL nuclear alignments in D. melanogaster (a), D. subobscura (b), D. baimaii (c), D. barbarae (d), D. birchii (e), D. bocki (f), D. diplacantha (g), D. fengkainensis(h), D. greeni (i), D. kikkawai (j), D. lacteicornis (k), D. leontia (l), D. mayri (m), D. ogumai (n), D. ohnishii (o), D. pectinifera (p), D. serrata (q), D. trapezifrons (r), D. truncata (s), D. tsacasi (t) and D. watanabei (u). Scale bars: 50 μm.

Figure 5

Species differences in sex-biased expression of Act79B, an MOL-enriched actin transcript. (a) Quantitative RT-PCR analysis. Genomic DNA (lanes 1 and 2) and first-strand cDNA (lanes 3–6) were prepared from tergites of abdominal segments A3–A6 of males (lanes 1, 3 and 5) and females (lanes 2, 4 and 6) of D. melanogaster, D. subobscura, and 6 species of the montium group (indicated below each panel) for Act79B (lanes 1–4) and 2 control protein genes, α-Tubulin (lane 5) and Act5C (lane 6). M: DL2000 DNA marker. Primers used are as shown in Table S3. (b–d,f–l) in situ hybridization analysis with probes for coding (b–d) or non-coding (f–l) sequences of the Act79B transcript. Act79B expression in abdominal muscles in the wild-type (b,f) and Act79B mutant (d,h) males and wild-type females (c,g) of D. melanogaster and in wild-type males (i,k) and females (j,l) of D. ogumai (i,j) and D. ohnishii (k,l). (e) Phalloidin staining reveals the MOL in a Act79B mutant male of D. melanogaster even though the mutant lacks Act79B expression (d). The true MOLs with Act79B hybridization signals are indicated with arrowheads. The tergite regions typically occupied by the MOL are circled with dotted lines. Oenocytes emit autofluorescence, resulting in a segmentally repeated labelling pattern marked with *. Scale bars: 200 μm (b,e).

Figure 6

Ancestral reconstruction of the neuromuscular sexual dimorphism in A5 for a Bayesian tree by BBM analysis. The presence or absence of the neuromuscular sexual dimorphism in the A5 segment was determined based on the results shown in Figs. 1 and S3 and Table 1. Pie charts along nodes indicate the probability of ancestral distribution of the neuromuscular sexual dimorphism estimated by BBM analysis (left-hand side panel). It is inferred that the neuromuscular sexual dimorphism was lost at the MRCAs of several lineages indicated by shading. The ancestral distribution of pre-mounting song states (right-hand side panel) is similarly constructed based on our previous observations (summarized in Fig. S5 according to Fig. 3 in Chen et al.) and our unpublished result. The pre-mounting song was probably lost in the two lineages indicated by shading.