Repeated evolution of asymmetric genitalia and right-sided mating behavior in the Drosophila nannoptera species group

Background Male genitals have repeatedly evolved left-right asymmetries, and the causes of such evolution remain unclear. The Drosophila nannoptera group contains four species, among which three exhibit left-right asymmetries of distinct genital organs. In the most studied species, Drosophila pachea, males display asymmetric genital lobes and they mate right-sided on top of the female. Copulation position of the other species is unknown. Results To assess whether the evolution of genital asymmetry could be linked to the evolution of one-sided mating, we examined phallus morphology and copulation position in D. pachea and closely related species. The phallus was found to be symmetric in all investigated species except D. pachea, which displays an asymmetric phallus with a right-sided gonopore, and D. acanthoptera, which harbors an asymmetrically bent phallus. In all examined species, males were found to position themselves symmetrically on top of the female, except in D. pachea and D. nannoptera, where males mated right-sided, in distinctive, species-specific positions. In addition, the copulation duration was found to be increased in the nannoptera group species compared to closely related outgroup species. Conclusion Our study shows that gains, and possibly losses, of asymmetry in genital morphology and mating position have evolved repeatedly in the nannoptera group. Current data does not allow us to conclude whether genital asymmetry has evolved in response to changes in mating position, or vice versa. Electronic supplementary material The online version of this article (10.1186/s12862-019-1434-z) contains supplementary material, which is available to authorized users.

perhaps independent [13,14] , or that behavior could simultaneously impede and drive 48 evolutionary diversification of different characters [12,15,16] . So far, it appears that the 49 effects of behavioral changes on the evolution of morphological traits cannot be 50 generalized and that they require case-specific assessments. 51 The evolution of left-right asymmetric genitalia in insects is a case where 52 morphology was proposed to have evolved in response to changes in mating behavior 53 [17] . Asymmetric genitalia are observed in many species and phylogenetic studies 54 indicate that they have evolved multiple times independently from symmetric ancestors 55 [18,19] . While most extant insect species copulate with the male being on top of the 56 female abdomen, the ancestral mating position in insects is inferred to be a configuration 57 with the female on top of the male [18,20,21] . The extant male-on-top configuration has 58 likely evolved multiple times in insects [20] . Such changes in mating position probably 59 altered the efficiency of male and female genital coupling, and may have led to the 60 evolution of genital asymmetries to optimize the coupling of genitalia [17] . 61 The nannoptera species group belongs to the genus Drosophila and consists of four 62 described species that feed and breed on rotten pouches of columnar cacti of the genus 63 found to be asymmetrically bent (n=10). Two asymmetric spurs were found at the ventral 113 apical tip of the aedeagus, with the right spur being consistently longer than the left spur 114 (Fig. 2e, Supplementary Fig.2). However, in contrast to D. pachea, no dorso-apical 115 gonopore was observed on the right side of the apex. Aedeagi of D. nannoptera males 116 (Fig. 2k, Supplementary Fig. 3) were found to be symmetric (n=15). The ventral side of the 117 apex revealed two apical elongations with slightly variable lengths at the left and right side 118 (n=15, Supplementary Fig.3). The variation in length was not directional and thus 119 considered to reflect random fluctuating asymmetry. The ventral tip of the aedeagus of D. 120 machalilla (atalaia species group) (n=10) displayed two lateral hooks (Fig. 2n, 121 Supplementary Fig.4), of the same length on both sides. The aedeagus of D. bromeliae 122 showed two lateral symmetric ridges (n=10) (Fig. 2q, Supplementary Fig. 5). In summary, 123 aedeagus asymmetry was only observed in D. pachea and D. acanthoptera, and distinct 124 phallus structures were found to be asymmetric in these species. 125

D. pachea and D. nannoptera males mate right-sided 126
The position of the male during copulation has not been described for any of the 127 closely related species of D. pachea. In this study, we assessed copulation postures in D. For each species, we introduced a single virgin female and a single virgin male into 142 a circular mating chamber and recorded the couple until copulation ended or for 45 min 143 when no copulation was detectable. We obtained 315 movies, of which 111 were used for 144 assessing courtship duration, 146 for copulation duration and 124 for copulation posture 145 analysis (supplementary dataset 3). Most movies were discarded because no copulation 146 occurred or individuals had damaged wings or legs (all reasons listed in supplementary 147 dataset 3). As previously described [36][37][38] , copulation duration varied significantly 148 among species (ANOVA, df1 = 9, df2 = 136, F = 73.38, p < 2e-16) ( Table 1). We could 149 reproduce a previously reported trend that copulation duration in nannoptera group 150 species was remarkably long compared to D. mojavensis and D. buzzatii of the repleta 151 group, with copulation duration of 88.49 min ± 35.18 min for D. acanthoptera, 29.58 ± 7.86 152 min for D. pachea and 11.9 ± 4.2 min for D. nannoptera (mean ± SD). In comparison, 153 copulation duration of D. buzzatii 1.79 ± 0.65 min and D. mojavensis 2.3 ± 0.35 min (mean 154 ± SD) of the repleta species group was shorter and similar to D. machalilla 2.28 ± 0.53 min 155 and D. bromeliae 0.92 ± 0.28 min (mean ± SD) ( Table 1). 156 To assess mating posture, we calculated the angle between a line drawn through 157 the male head midline and the female scutellum tip and a second line drawn through the 158 female head midline and the female scutellum tip ( Supplementary Fig. 7A). The angle was 159 set positive when male head lies on the right side of the female and negative when on the 160 left. The camera view relative to the fly couple position within the mating cell may affect the 161 measured angle in each experiment but the sign of the average mating angle taken from 162 different recordings for each species should accurately reflect the one-sidedness of the 163 male mating position. As a consequence, we expected a one-sided copulation position to 164 produce a consistent positive or negative distribution of angle values, while symmetric 165 mating positions should result in an angle distribution around zero. 166 To compare mating angles between species, it is necessary to examine copulation 167 postures at the same corresponding time point during copulation. At copulation start, the 168 male position on top of the female was found to be greatly variable between couples, even 169 within a single species, so this time point was not considered appropriate for our 170 comparative analysis. Since copulation duration varies greatly between species, finding 171 another comparable time point across species was not trivial. We subdivided copulation 172 into two phases, an initial phase where the male is on top of the female abdomen but 173 consistently moving legs and abdomen, and a second phase when the male maintains an 174 invariant position relative to the female, which can sometimes walk or move its legs 175 ( Supplementary Fig. 6). The "settling time point" is defined as the time point between the 176 first and second phase, when the male adopts an invariant position relative to the female. 177 For our cross-species analysis we chose to assess copulation angle at two time points: (1) 178 right after the male had settled into an initial invariant copulation position (the settling time 179 point) and (2) at 10% of elapsed time between the settling time point and the end of 180 copulation (10% stable copulation time point). For species with a mean copulation duration 181 > 2.5 min, > 15 min or > 60 min, we also measured the angles every 2.5 min, 5 min or 10 182 min, respectively. This allowed us to follow mating postures of each species over the 183 course of copulation. 184  Table 2). No significant one-sided copulation postures were detected in D. 187 acanthoptera and the other seven tested species including D. melanogaster (Fig. 3a,b). 188 Over the course of copulation, mating angles continued to range over zero for D. 189 melanogaster and D. acanthoptera (Fig. 3c,d)  Left-sided angles were only observed during the first two minutes of copulation. On 210 average, the male tended to initially adopt a right-sided copulation posture with an angle of 211 10.36° ± 6.88° (mean ± SD) (n=25) between 0-1 min after copulation start (Table 3). Over 212 the course of copulation, the angle then increased to 27.16° ± 10.81° (n=29) between 3-4 213 min after copulation start (Table 3) left and right sides. Here, we compared aedeagus morphology of at least 10 specimens of 232 five different species that belong to the nannoptera species group and closely related 233 species. We did not detect aedeagus asymmetry in the tested species outside of the 234 nannoptera species group and found that within the nannoptera group only D. 235 acanthoptera and D. pachea but not D. nannoptera reveal striking left-right asymmetries 236 ( Fig. 4). We did not evaluate aedeagus asymmetry of D. wassermani, as this species is not 237 available for examination and our attempts to catch specimen in the wild were not 238 successful (see materials and methods). Asymmetries differed between D. pachea and D.

Long copulation duration is specific to the nannoptera group species 256
We observed that nannoptera species copulated considerably longer than any 257 representative species of the close outgroup lineages (Fig. 4). This trend was previously 258 reported by Pitnick and Markow (1991) [36,37] , where the authors compared 259 copulation duration of nannoptera group species with repleta group and other species. 260 Here we included two additional closely related species, D. machalilla and D. bromeliae, 261 and observed that their copulation durations were relatively short. Our observations 262 therefore indicate that a long copulation duration is specific to the nannoptera group. nannoptera. Our previous data from D. pachea [26,29] was re-analysed in this study 283 with a different measurement approach and led to the same conclusion as our earlier 284 reports. In addition, we found that D. nannoptera adopts a right-sided mating position with 285 angle values that were slightly higher than in D. pachea (Fig. 3) Across the nannoptera group, we find no striking correspondence between right-307 sided mating posture and asymmetric male genitalia. For example, D. acanthoptera has an 308 asymmetric aedeagus but mates in a symmetric overall posture. On the opposite, no 309 directional asymmetry is detected in the male (external and internal) genitalia of D. 310 nannoptera, but males adopt a right-sided copulation posture (Fig. 4). Based on our 311 phylogeny, D. nannoptera presents the earliest branching lineage within the nannoptera 312 group. In this sense, right-sided mating postures could have originated earlier during 313 evolution than asymmetric morphologies and may have been lost in D. acanthoptera. 314 However, the internode branch length between the split of the D. nannoptera lineage and 315 the separation of D. acanthoptera and D. pachea is short and statistical support is weak 316 [28] . Thus, phylogenetic relationships within the nannoptera group remain to be 317 resolved and it is more appropriate to regard all nannoptera species as sister species. 318 So far, we conclude that both right-sided copulation behavior and asymmetric male 319 genitalia evolved within the nannoptera species group and that diversification of both traits  (Fig. 4). The association of right-sided mating with giant sperm 331 production actually holds better than with asymmetric male genital morphology because D. 332 acanthoptera has an asymmetric aedeagus but has relatively small sperm [48] and 333 mates in a symmetric overall posture (Fig. 4)

Analysis of aedeagus asymmetry by light microscopy 407
The terminal segments of the male abdomen were picked out with fine forceps and 408 boiled for 10 min in two drops of 30% KOH. Genital parts were further dissected on a 409 The divergence estimate for all analyzed species was set to 40 ± 5 Ma [56] . 430

Copulation recording 432
Emerged flies (0-14 h) were anesthetized with CO 2 , separated according to sex and 433 transferred into food vials in groups of either 5 females or 5 males using a Stemi 2000 434 (Zeiss) stereo microscope and a CO 2 -pad (Inject+Matic sleeper). Flies were maintained at 435 22°C or 25°C until they reached sexual maturity (Supplementary Table 1 Supplementary Fig. 6). The remaining copulation period was 470 defined as the stable copulation period (Supplementary Fig. 6). In fact, this period was between the third thoracic and first abdominal segment. Position landmarks were placed 503 manually on each image using imageJ and data analysis was done using R. Briefly, 504 coordinates (supplementary dataset 4) were rotated and scaled, so that all P1 points were 505 superimposed and all P2 points as well (supplementary Fig. 7B-K). The angle P1-P2-P3 506 ( Supplementary Fig. 7A) was used to measure one-sidedness of mating positions (Fig. 3). 507 Repeatability of landmark positioning was assessed by two independent rounds of 508 coordinate acquisition for all species at one specific time point during copulation, the 10% 509 Flies were reared and isolated before copulation as described above. One female 519 and one male were CO 2 anesthetized and transferred onto a white plastic support (mating 520 cap) and were caged with a transparent plastic cylindrical 25 mm x 7 mm cap. Once 521 courtship was observed, mating caps were put on a motorized horizontally turning stage 522 considerations mentioned for these species according to EU Directive 86/609-STE123. 542

Consent for publication 543
"Not applicable" 544

Availability of data and material 545
The data sets supporting the results of this article will be made available in the DRYAD 546 repository. 547 548

Competing interests 549
The authors declare no competing financial interests 550