Asymmetric and Spiraled Genitalia Coevolve with Unique Lateralized Mating Behavior

Asymmetric genitalia and lateralized mating behaviors occur in several taxa, yet whether asymmetric morphology in one sex correlates or coevolves with lateralized mating behavior in the other sex remains largely unexplored. While lateralized mating behaviors are taxonomically widespread, among mammals they are only known in the harbor porpoise (Phocoena phocoena). Males attempt copulation by approaching a female exclusively on her left side. To understand if this unusual lateralized behavior may have coevolved with genital morphology, we quantified the shape of female and male harbor porpoise reproductive tracts using 2D geometric morphometrics and 3D models of the vaginal lumen and inflated distal penis. We found that the vaginas varied individually in shape and that the vaginas demonstrated both significant directional and fluctuating asymmetry. This asymmetry resulted from complex 3D spirals and vaginal folds with deep recesses, which may curtail the depth or direction of penile penetration and/or semen movement. The asymmetric shapes of the vaginal lumen and penis tip were both left-canted with similar angular bends that mirrored one another and correspond with the left lateral mating approach. We suggest that the reproductive anatomy of both sexes and their lateral mating behavior coevolved.

www.nature.com/scientificreports www.nature.com/scientificreports/ hypotheses 16,17 . Additionally, among mammals, asymmetric penises are most common in cetartiodactyls 4 , which have a fibroelastic penis that maintains a turgid state and is resistant to bending 18 . Both male and female genital morphology in cetaceans is particularly complex. The penis tip of cetaceans is pliable and may be under voluntary control 19,20 . Female harbor porpoises (Phocoena phocoena) have one of the most complex vaginal morphologies and the largest number of vaginal folds described in any cetacean species 16 . Male harbor porpoises have a large blunt knob at the terminal end of the relatively long penis shaft (among the proportionally longest in cetaceans), which connects to a thin distal tip that bends away from the midline 19 . During copulation, the blunt knob may be obstructed from penetrating the cranial vagina by the largest vaginal fold, as suggested by ex vivo testing 21 . The thin distal penis tip has been hypothesized to enable passage between the remaining cranial vaginal folds to deposit sperm close to the cervix 21 . The vaginal folds may also dampen biomechanical forces to surrounding tissues during copulation 22 . The close biomechanical interaction and fit between male and female genitalia in general during copulation suggests that reproductive morphology may coevolve 15 .
Harbor porpoises are currently the only known mammal with lateralized mating behavior 7 . Although asymmetries in marine mammal behavior tend to be right-biased (e.g., feeding 23 ), recent documentation of 142 copulation attempts by male harbor porpoises in San Francisco Bay showed they were all performed on the female's left side 7 . Copulatory attempts consisted of rapid (1-4 seconds), high-energy, precision-timed approaches, during which the males often breached out of the water with their penis fully extended and attempted to drive their penis into the female's vaginal opening 7 . The complexity of their genital morphology and their lateralized mating behavior make the harbor porpoise an exemplary taxon to explore the potential relationship between reproductive morphology and mating behavior. To characterize and quantify genital shape and assess the influence of asymmetry on overall genital shape, we used two-dimensional geometric morphometrics (2DGM) and three-dimensional models of male and female harbor porpoise genitalia. Asymmetry has been previously reported in female genitalia in a wide range of taxa, but can be inconspicuous when present [3][4][5]13,24 . Therefore, we examined the pattern of asymmetry in vaginal folds in both 2D and 3D by generating models of the vaginal lumen, in addition to assessing gross morphology. This is the first study to address the question of whether lateralized mating behaviors affect genital morphology in mammals, and allows for a better understanding of the drivers of sexual patterns in both cetaceans and the wide variety of animals that have asymmetric genitalia with potentially previously unrecognized lateralized mating behaviors.

Methods
Female (n = 10) and male (n = 3) reproductive tracts were obtained from fresh or moderately decomposed deceased sexually mature harbor porpoises that died of natural causes or fisheries interactions in the San Francisco Bay Area. Specimens were frozen and shipped from The Marine Mammal Center to necropsy facilities at either Mount Holyoke College or the National Oceanic and Atmospheric Administration's Southwest Fisheries Science Center. Specimens were obtained under a National Marine Fisheries Service salvage permit to D.N.O. or S.L.M. Female reproductive tracts consisted of the external genital opening through to the uterine horns and ovaries 25 . Male reproductive tracts included the entire penis tip and shaft to the pelvic bones. Sperm was found in the testes of all male specimens, confirming their sexual maturity at the time of death. Specimen information (i.e., body length, weight, and female reproductive state) was provided by The Marine Mammal Center (Table 1).
Nine female reproductive tracts were dissected in dorsal recumbency (the specimen lying on its dorsal side with the ventral surface exposed). The reproductive tract is oriented horizontally (cranial-caudal) in the female's body. As vaginal folds are most prominent on the dorsal side of the vagina 25 , incisions were made on the ventral side to visualize the shape of internal structures 21 . A single incision was made down the ventral midline of each reproductive tract, from the internal bifurcation of the uterine horn through the clitoris 25   midline of the reproductive tract, with the long axis of the vaginal canal oriented such that the vaginal opening was at the bottom of the image. Scales were positioned in both transverse and coronal (frontal) planes. Two-dimensional geometric morphometrics were employed to quantify variation in the shape of the cervix, first large cranial vaginal fold, large main vaginal fold, and vaginal entrance. The cranial and caudal bounds of each region were outlined using both landmarks (n = 16) and semi-landmark curves (n = 20 curves, 240 semi-landmarks) in TPSDig2 (Fig. 1a) 26 . All landmarks were positioned by one author (D.N.O.) to eliminate inter-observer error 5,17,27,28 . As positioning of landmarks in soft tissue can be challenging, they were applied to each image three independent times to quantify intra-observer measurement error. Landmark configurations were imported into R 29 , where a Generalized Procrustes Analysis (GPA) was performed to rescale, translate, and rotate all landmark configurations into the same shape space 30 using the geomorph package 31 . Semi-landmark curves were slid to minimize the bending energy between semi-landmarks 32 . Landmark configurations were then subjected to principal component analysis (PCA) to visualize the major axes of shape change among the specimens and to ensure that measurement error triplicates plotted more closely to one another than to other individuals. Although total length did not vary substantially within the harbor porpoises (146-174 cm), the effect of size on the shape data was tested using the Regression Score metric with log-transformed total length as a proxy for the size of each individual 33,34 .
To quantify 2D asymmetry, a multi-factor ANOVA was used with individuals (symmetric variation), sides (directional asymmetry), and the interaction between individuals and sides (fluctuating asymmetry) as factors 35,36 . Directional asymmetry captures asymmetry where one side is consistently shaped differently from the other (e.g., fiddler crab claws 37 ). Fluctuating asymmetry captures random right-left perturbations from symmetry that are a result of developmental or environmental factors 36 . The proportion of total variation explained by each individual factor was calculated using the η 2 effect size metric 38 .
To capture 3D asymmetry of the vaginal canal, endocasts were made of the lumens prior to dissections. One endocast was made from a female collected from San Francisco Bay and four endocasts were made from adult females obtained from other populations along the Pacific coast of the United States. Female reproductive tracts were suspended and filled with Mold Star ® 16 FAST silicone. The endocasts were carefully removed to prevent damage to the molds or tissue and to identify the corresponding ventral and dorsal sides of the female. 3D models of the endocasts were then generated with photogrammetry. The complete circumference of each endocast was photographed using a Canon EOS Rebel T5i camera with a 100 mm lens and was illuminated with four LED lights. Three-dimensional models were reconstructed in 3DF Zephyr lite (3Dflow SRL, Verona, Italy) and scaled and bisected in sagittal and lateral planes using Autodesk ® ReCap Photo (v. 22  The different planes were qualitatively compared to determine the degree of symmetry. The pathway from the vaginal opening to the cervix was digitally superimposed onto the image (Fig. 2b).
The penis tips and shafts of three male harbor porpoises were filled to distention with melted Vaseline (Unilever) to recreate the likely shape attained during erection (Table 1). Vaseline was injected into the corpora cavernosa and corpora spongiosa tissues with an 18-gauge needle connected to a syringe until turgidity and resistance to further distension was achieved (e.g. 21,39 ). The artificially inflated penises were engorged to similar dimensions as observed in several photographs of free-swimming harbor porpoises with erections (Fig. 2d) 7 . 3D models were made using photogrammetry with the same methods described above for the vaginal endocasts.
The digital penis models were angled in a consistent coronal plane with the tip facing cranially using Autodesk ® ReCap Photo. A black line was drawn down the dorsal midline of each 3D penis model (from the attachment of the retractor muscle to the distal tip) to calculate the angles of each bend in the penis tip relative to the midline www.nature.com/scientificreports www.nature.com/scientificreports/ which was set at 0 degrees (Fig. 2a). Major changes in penis angles were calculated relative to the preceding angles using the angle tool in ImageJ (v.1.52k 40 ).
One female reproductive tract was dissected in a ventral recumbency (the specimen lying on its ventral side with the dorsal surface exposed) with an incision made down the dorsal midline (Table 1; ID C-441). This orientation reflects the natural position of females in the wild (ventral side down) during male left-sided sexual approaches 7 . The pathway from the vaginal opening to the cervix around the complex vaginal folds was digitally superimposed onto an image of this dissected reproductive tract (Fig. 2c). We then qualitatively compared the superimposed pathway through the vaginal lumen with the shape of the inflated penises (Fig. 2a).
Significance statement. We explored the coevolution of genital morphology and mating behavior asymmetry between male and female harbor porpoises, the only mammal known to demonstrate exclusive lateralized mating behavior. By comparing 2D and 3D shapes of excised male and female reproductive tracts, we found that genital asymmetry is likely functional. Male and female genital shapes are congruent in angular bends and a  Table 2. Angles of the inflated penis tips of three harbor porpoises. The first angle is relative to the long axis of the penis shaft such that 0° angles parallel to shaft in the cranial direction and the subsequent angles are relative to the preceeding angle. The four angles show marked directional changes in the penis tip (Fig. 3a). Prior to measurement, penises were angled in a coronal plane with the tip facing cranially. left-sided sexual approach by males is obligatory to overcome physical obstructions inside the female reproductive tract. Our results suggest male copulatory behavior and morphology conform with female morphological modifications to achieve fertilization in a system which may undergo an evolutionary arms race.

Results
We found substantial variation in vaginal shape in 2D using PCA. Principal component (PC) 1 (26.06% of total shape variation) was driven by the relative size of the lower vagina (caudal to the largest vaginal fold; Fig. 3a), PC2 (21.96% of total shape variation) was driven by a left-or right-biased curvature of the entire reproductive tract (Fig. 3a), and PC3 (18.6% of total shape variation) was driven by the relative widths of the cervix and vagina (Fig. 3b,c). Vaginal shape was not significantly correlated with total body length (p = 0.757; Fig. 3d). Based on ANOVA effect sizes, individual variation accounted for 52.6% of total shape variation, directional asymmetry for 8.3% of total shape variation, and fluctuating asymmetry for 37.1% of total shape variation. All three of these factors were significant (p < 0.001; Fig. 3e). Measurement error, by contrast, only comprised 2% of the total shape variation, demonstrating that the landmarking process did not generate a substantial error component. The internal vaginal lumen is highly asymmetric due to both complex 3D spirals and vaginal folds with deep recesses (Fig. 1b,c). The largest vaginal fold had a mean thickness of 19.6 mm (S.D. = 4.7 mm, n = 8). The pathway through the vaginal lumen to the cervix had an overall bend and spiraled to the left in all five endocasts when positioned in ventral recumbency, as found during mating under natural conditions (Fig. 2b). This bend, and two additional moderate bends in the vaginal lumen, are generated by the vaginal folds. When viewed dorsally, the tips of all three inflated penises originate on the left side of the blunt knob and cant to the left in a series of four consecutive bends ( Table 2; Fig. 2a). Specifically, relative to the midline of the long axis of the shaft, the penis angles marginally to the right slightly proximal to the blunt knob, then bends ~90° to the left and downwards toward the shaft, followed by subtler bends to the right then left near the distal tip. These bends parallel the shape of the internal vaginal lumen (Figs. 1c and 2b) and accurately reflect the penis shape of free-swimming harbor porpoises in San Francisco Bay (Fig. 2d) when males attempt copulation. A video of a copulatory attempt shows the body orientation of males relative to females (Supplemental Video 1).

Discussion
We provide evidence that genital asymmetry has coevolved with lateralized mating behavior in harbor porpoises, the first example amongst mammals. The vaginal lumen of harbor porpoises is significantly directionally asymmetric, and its shape corresponds to the leftward cant of the erect penis tip. The left-sided mating behavior of male harbor porpoises appears to have coevolved with genital shape variation in both sexes so that the penis can circumvent large and protruding vaginal folds 21 . The close genital correspondence could be the result of several potential coevolutionary mechanisms, including sexual conflict, female choice, and "lock-and-key". As hybridization is common among cetaceans 41 , it is unlikely a "lock-and-key" mechanism, and the close correspondence likely arises as a result of an intersexual evolutionary arms race in harbor porpoises. In the San Francisco Bay population, males initiate all mating events, attempt copulations throughout the year, do not engage in pre-copulatory courtship, and approach females in all reproductive states 7 . Females may respond to the constant pursuit of males by allowing copulations that reduce costs of evasion and harassment. However, they have also evolved sophisticated mechanisms that may control paternity including asymmetric genitalia, complex vaginal folds, changes in body positioning during copulation that may function to misalign the optimal angle of penetration and prevent the ejaculate from reaching the cervix, and through resistant behaviors (e.g., dives 42 ). Both sexes appear to exhibit both behavioral and anatomical adaptations and counter-adaptations that may control paternity, although it is possible that an evolutionary arms race has not reached an equilibrium end point (e.g. 43 ). It is clear that penile penetration into the most cranial regions of the vagina requires a specific angle of entry to bypass the vaginal folds. By angling their bodies slightly when males approach, females could potentially prevent the tip of the penis from penetrating past their vaginal folds 21 . Males seem to have evolved complex penile morphology with numerous angles that match the internal shape of the vagina and may increase their chances of successful copulation. The large relative testes size of harbor porpoises compared to other cetaceans suggests a substantial investment in post-copulatory sperm competition 44,45 and there is likely strong selection on females to evolve mechanisms to control paternity after insemination by multiple males.
Although all the female reproductive tracts examined were asymmetric and complex, we found that vaginal shape strongly varied among individuals. Female genitalia were previously considered to be less variable than male genitalia 46 , yet emerging patterns suggest a high level of intra-and inter-specific variation in females, which highlights the need to continue to explore female genital morphological variation across taxa 47 . Reproductive state and history could have contributed to this variation, but our sample size was too small to examine this possibility quantitatively. For example, a study on reproductive tract lengths in common bottlenose dolphins (Tursiops truncatus) found that pregnant females had longer vaginal lengths than lactating or resting adult females, although the sample size was also too small for statistical analysis 25 . We qualitatively note that pregnant harbor porpoises generally had more symmetrical cervices, potentially reflecting stretching, while resting females had more asymmetrical cervices. Recent assessment of soft vaginal structures in other taxa using 2DGM have similarly found that vaginal shape is highly variable and influenced by factors such as ontogeny and reproductive state (beetles 48 ; snakes 49 ; cetaceans 17 ; sharks 5 ). However, much of the documented variation remains unexplained.
More than 8% of total shape variation in the harbor porpoise vagina was explained by directional asymmetry. As directional asymmetry is often functional 50,51 and all vaginal endocasts canted to the left, the data support a left-sided constraint on males to approach females only from one direction and body orientation. We found that fluctuating asymmetry was significant and explained 37% of total variation. This high degree of fluctuating asymmetry in soft tissue structures is common (e.g., dogfish shark vagina 5 ), and may result from less canalization in comparison with rigid hard tissues 52,53 . While genital laterality is to the left, other morphological asymmetries known in (2020) 10:3257 | https://doi.org/10.1038/s41598-020-60287-w www.nature.com/scientificreports www.nature.com/scientificreports/ harbor porpoises are to the right (e.g., flipper size 54 ; nasal complex asymmetries 55,56 ), and right-dominated asymmetries in non-mating behavior occur in other marine mammals 23 . This makes it likely that the genital asymmetry is adaptive under the influence of selection, and does not result from pleiotropy in harbor porpoises, which could favor right-sided development concurrently with other right-sided morphological asymmetries.
We have demonstrated the development and coevolution of lateralized sexual behavior and genital morphology in harbor porpoises, which appears to reflect an intersexual evolutionary arms race. It is not yet possible to determine whether the lateralized behavior evolved in response to morphological evolution or vice versa, or how it became fixed in the species. Substantial future work on additional species is necessary to establish the commonality of correlations between lateralized behavior and morphology among vertebrates.