Evolution of sexual size dimorphism in the wing musculature of Drosophila

Fly stocks

Flies were reared on standard cornmeal and molasses food at room temperature. Most stocks of species other than D. melanogaster were obtained from the Drosophila Species Stock Center (now located at Cornell University, College of Agriculture and Life Science).

Measurement of wing muscle volume

To visualize and measure the volumes of wing muscles, hemithoraces from males or females of each species were dissected in PBS and fixed in 4% paraformaldehyde (buffered in PBS) for approximately 50 min at room temperature. To effectively stain the control wing muscles lining the lateral wall of the thorax, we removed the six large fibers of the dorsal longitudinal muscle after fixation. Fixed tissues were washed in PBS-TX (PBS with 1% Triton X-100), placed in PBS-TX containing Texas Red-X phalloidin (1:50; Life Technologies), and incubated for 3 to 4 days at 4 °C. Tissues were subsequently washed all day in PBS-TX, placed onto poly-lysine-coated coverslips, dehydrated through an ethanol series, cleared in xylenes, and mounted in DPX (Sigma-Aldrich). Tissues were imaged on a Zeiss LSM 510 confocal microscope at 10X with optical sections at 1 µm intervals. Confocal stacks of phalloidin-stained hemithoraces were imported into Amira (Thermo Scientific). The wing muscles were segmented and reconstructed by selecting and assigning pixels through the confocal series to labels of their respective wing muscle. Amira was used to measure muscle volume using the appropriate voxel dimensions (in mm).

Given species variation in overall body size between males and females, we measured a ‘normalized’ hg1, hg2, hg3, and hg4 muscle volume for each dissected hemithorax. The raw muscle volume of each hg muscle, the first basalare muscle, and the dorsal and ventral Tergopleural muscles were measured for each hemithorax. A normalized muscle volume was obtained by dividing the volume of each hg muscle for each hemithorax by the total muscle volume (i.e., hg1 + hg2 + hg3 + hg4 + b1 + d-Tp + v-Tp) of the respective hemithorax. The normalized muscle volumes for males and females of each species were used to calculate the Cohen’s d effect size score (Cohen, 1988) to estimate the magnitude of the difference in normalized muscle volumes between males and females. The effect size difference between males and females for each species was used to create a categorical dataset, categorizing each value as either “male enlarged,” “female enlarged,” or “monomorphic.” Species with a Cohen’s d ≥ 1.5 were considered to be dimorphic, whereas species with a Cohen’s d <1.5 were categorized as monomorphic. A Cohen’s d ≥ 1.5 means that the mean normalized muscle volume of one sex is above the 90th percentile of the opposite sex’s normalized muscle volume distribution.

Phylogenetic estimation

The phylogenetic estimate for the genus Drosophila utilizes sequence data available in the GenBank Nucleotide Database (Clark et al., 2016) for six nuclear genes (ADH, amyrel, Ddc, Gphd, Sod, and XDH; Tables S1 and S2). Phylogenetic relationships were estimated using taxon sampling similar to that of described previously (Russo et al., 2013) for the D. montium, D. melanogaster, and D. ananassae subgroups. The phylogeny was estimated using maximum likelihood in IQ-TREE v. 1.6.1 (Nguyen et al., 2015) from a concatenated alignment of the six genes. Each gene was partitioned by codon position to estimate models of sequence evolution using ModelFinder (Kalyaanamoorthy et al., 2017) in IQ-TREE, substitution models were chosen based on the Bayesian Information Criterion (BIC). The IQ-TREE analysis was run with 1000 ultrafast bootstrap replicates (Minh, Nguyen & Von Haeseler, 2013). Subsequent to estimation of the phylogeny the tree was pruned to include only taxa for which muscle volume data was available for the ancestral state reconstruction.

The phylogenetic analysis resulted in a well-supported phylogeny with relationships matching those found in previous studies (e.g., Russo, Takezaki & Nei, 1995). The backbone of the tree was largely supported, however some of the groups are not well supported in the maximum likelihood tree (Figs. S1, S2). As the majority of the backbone relationships were well supported, we pruned the phylogeny to include only species for which we have muscle volume data. The pruned tree is largely supported with the only unsupported nodes falling within the melanogaster species group (Fig. S3). While not all nodes within the melanogaster species group are well supported, the relationships recovered match previous studies of the Drosophila phylogeny.

Ancestral state reconstruction

An ancestral state reconstruction on the categorical dataset was run in R-Studio (R Core Team, 2017) using the ancestral character estimation tool in the package ‘ape’ (Paradis & Schliep, 2019). This function uses maximum likelihood estimation in a two-pass manner to work from the values at each tip back to the root of the tree, then working from the root back to the tips to determine the likelihood of each state at each node. Likelihood values for each node were then mapped onto the phylogenetic tree in pie chart format to indicate the relative likelihood of dimorphic (male-enlarged or female-enlarged) vs monomorphic being the ancestral state at that specific node.

The evolutionary history of hg1′s sexual size dimorphism

Given the role of hg1 and its size dimorphism in patterning D. melanogaster courtship song (Shirangi, Stern & Truman, 2013), we sought to determine the history of hg1′s sexual dimorphism in the Drosophila phylogeny: when did hg1′s size dimorphism emerge in Drosophila, and how stable has it been since it arose? We thus analyzed the volume of the hg1 muscle from males and females of 16 species within the melanogaster species group, which emerged ∼25 Mya (Russo, Takezaki & Nei, 1995), and 3 outgroup species (D. obscura, D. willistoni, and D. virilis). Using these species, we estimated a phylogeny using maximum likelihood methods, identified the species with a dimorphically or monomorphically sized hg1 muscle, then employed methods in ancestral state reconstruction to estimate the probability of a dimorphic or monomorphic hg1 muscle in the ancestral nodes of the phylogeny.

Overall body size typically differs between Drosophila males and females, and the magnitude by which males and females differ in size varies across species. We thus measured the relative hg1 muscle volume by normalizing the raw hg1 volume of each fly to the total volume of seven wing muscles. To determine if a species has a dimorphically sized hg1 muscle, we measured the magnitude, i.e., Cohen’s d effect size (Cohen, 1988), of the difference in normalized hg1 volume between males and females of each species (Figs. 2 and 3). A species was categorized as “dimorphic” if the male’s mean hg1 volume fell above the 90th percentile of the female’s hg1 muscle volume distribution. Such species have a Cohen’s d value of 1.5 or greater (Cohen, 1988). Species with a Cohen’s d value below 1.5 were deemed “monomorphic.”

The melanogaster species group consists of several subgroups that form three main clades: the ananassae species subgroup, the montium species subgroup, and the Oriental lineage, which includes the melanogaster, suzukii, eugracilis, and elegans subgroups. Ancestral state reconstruction of hg1′s sexual size dimorphism (Fig. 3, Fig. S1, Table S1) indicates that a sexually dimorphic, male-enlarged hg1 muscle was most likely present in the ancestor of the melanogaster, suzukii, and eugracilis subgroups (probability of a dimorphic ancestor = 0.839). Ancestral nodes within the melanogaster, suzukii, and eugracilis subgroups remain strongly dimorphic (probability of dimorphic ancestor ≥ 0.859), indicating that the size dimorphism was lost independently twice in D. eugracilis and D. orena, both of which have a monomorphic hg1 muscle. Nearly all species studied (8/9 species) outside of the melanogaster, suzukii, and eugracilis subgroups have a monomorphically sized hg1 muscle, suggesting that the dimorphism may have first emerged in the ancestor of the melanogaster, suzukii, and eugracilis subgroups. However, given that D. fuyamai has a strongly dimorphic hg1 muscle, it is possible that a dimorphic hg1 muscle emerged earlier in the phylogeny in the ancestor of the Oriental lineage, or independently in the lineage leading to D. fuyamai. Nevertheless, our results indicate that a sexually dimorphic hg1 muscle arose within the Oriental lineage and was subsequently lost at least twice independently. Moreover, it appears that the relative position and shape of the hg1 muscle within the thoracic box are largely conserved across the phylogeny (Fig. S4).

Sexual size dimorphisms in the Drosophila wing musculature are not limited to hg1

The evolutionary transitions in hg1 described above may have been associated with changes in male courtship song (e.g., gain or loss of sine song) or other sexually dimorphic wing-based behaviors. Alternatively, evolutionary transformations in the sexual size of hg1 may have been accompanied by compensatory changes in other wing muscles that would abrogate evolutionary changes in behavior (e.g., if the behavior was under stabilizing selection). There is precedent for this hypothesis, as hg1 is one of four control wing muscles (hg1–4) that insert into a common tendon that influences the mechanical properties of the posterior notal wing process at the wing hinge (Fig. 1); moreover, all four hg muscles are active during D. melanogaster courtship song (O’Sullivan et al., 2018). Given the potential functional overlap among the hg muscles, loss of the size dimorphism in hg1 may have been associated with a compensatory gain of a size dimorphism in another hg wing muscle, or vice versa. To test this hypothesis, we analyzed the evolutionary history of sexual size dimorphism in the hg2, hg3, and hg4 muscles.

Analyses of hg2, hg3, and hg4 muscle size from males and females of the 19 species studied above revealed that the hg2 and hg3 muscles evolved a size dimorphism multiple independent times in the Drosophila phylogeny (Figs. 4A, 4B, Fig. S1 Table S4). Hg4, however, has remained anatomically monomorphic through Drosophila’s evolutionary history (Fig. 4C). The magnitude of the size dimorphisms in hg2 and hg3 are generally lower than those observed in hg1 (Fig. 2). Notably, if the gain or loss of a size dimorphism in hg1 (Fig. 3) was compensatory to an anatomical transition in another hg muscle, we would expect to find either hg2, hg3, or hg4 to be male-enlarged where hg1 is monomorphic or vice versa. However, none of the hg muscles display such a pattern, indicating that evolutionary changes in hg1 were most likely not compensatory to changes in other hg wing muscles.

Our analyses reveal a surprising amount of species variation in sexual size dimorphism in the Drosophila wing musculature. Ancestral state reconstruction of hg2′s sexual size dimorphism (Fig. 4A) indicates that the ancestor of the Oriental lineage most likely had a monomorphically sized hg2 muscle (probability of a monomorphic ancestor = 0.999). The hg2 muscle appears to have subsequently evolved a size dimorphism in members of the suzukii species subgroup (i.e., D. mimetica, D. prostipennis, and D. biarmipes). Outside of the Oriental lineage, hg2 evolved a size dimorphism at least once in the ananassae species group. It is noteworthy that unlike the dimorphisms in hg1, which were always male-enlarged, hg2 has evolved male- and female-enlarged size dimorphisms.

Ancestral state reconstruction of hg3 and hg4′s sexual size dimorphism (Fig. 4B, Fig. 4C) reveals that a monomorphic hg3 and hg4 muscle was most likely present throughout the basal lineages and the Oriental lineage. A size dimorphism in hg3 evolved independently four times in D. elegans, D. yakuba and D. prostipennis (male-enlarged), and in D. erecta (female-enlarged) (Fig. 4B). None of the taxa sampled display a dimorphic hg4 muscle (Fig. 4C). Like hg1, the relative position and shape of the hg2–4 muscles within the thorax have not changed in any obvious way (Fig. S4).