Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Nov 1.
Published in final edited form as: Evolution. 2014 Oct 20;68(11):3296–3306. doi: 10.1111/evo.12516

Sexual selection targets cetacean pelvic bones

J P Dines 1,2,*, E Otárola-Castillo 3,4, P Ralph 5, J Alas 5,6, T Daley 5,7, A D Smith 5, M D Dean 5,*
PMCID: PMC4213350  NIHMSID: NIHMS626667  PMID: 25186496

Abstract

Male genitalia evolve rapidly, probably as a result of sexual selection. Whether this pattern extends to the internal infrastructure that influences genital movements remains unknown. Cetaceans (whales and dolphins) offer a unique opportunity to test this hypothesis: since evolving from land-dwelling ancestors, they lost external hind limbs and evolved a highly reduced pelvis which seems to serve no other function except to anchor muscles that maneuver the penis. Here we create a novel morphometric pipeline to analyze the size and shape evolution of pelvic bones from 130 individuals (29 species) in the context of inferred mating system. We present two main findings: 1) males from species with relatively intense sexual selection (inferred by relative testes size) have evolved relatively large penises and pelvic bones compared to their body size, and 2) pelvic bone shape diverges more quickly in species pairs that have diverged in inferred mating system. Neither pattern was observed in the anterior-most pair of vertebral ribs, which served as a negative control. This study provides evidence that sexual selection can affect internal anatomy that controls male genitalia. These important functions may explain why cetacean pelvic bones have not been lost through evolutionary time.

Introduction

The rapid divergence of male genitalia, probably the result of sexual selection affecting male-male competition and/or male-female interactions (Eberhard 1985; Dixson 1998; Hosken and Stockley 2004; Miller 2010), has emerged as a preeminent pattern in evolutionary biology. Male genitals may have evolved to remove sperm or otherwise reduce the fertility of competing males, to induce the female to accept insemination, to harm the female and inhibit her from re-mating, and/or to sneak matings (reviewed in Eberhard 1996; Simmons 2001; Arnqvist and Rowe 2005). Thus, male genitals are not solely involved in transfer of gametes, but participate in various arenas of competition and conflict, the intensity of which is expected to intensify in relatively promiscuous species. Accordingly, male genitalia diverge more rapidly in more promiscuous species (Arnqvist 1998; Hosken and Stockley 2004; Ramm 2007), and some genital shapes are more effective at securing reproductive fitness under more competitive contexts (House and Simmons 2003, 2005; Stockley et al. 2013; Simmons and Firman 2014). One underlying mechanism for the divergence of male genitalia may be coevolution with the female. Over evolutionary time, selection may favor females that morphologically or behaviorally inhibit insemination, possibly as an indirect means to select the fittest mates from the population. Selection may also favor males that counteract these measures, leading to a coevolutionary conflict of interest that drives divergence in both male and female reproductive anatomy (Baumgardner et al. 1982; Higginson et al. 2012). With increased intensity of sexual conflict,

In addition to genital morphology per se, selective processes could affect the way that males move their genitals in the events leading up to and including copulation, but this hypothesis remains largely unexplored. The pelvic bones of cetaceans (whales and dolphins) offer a unique opportunity to test this hypothesis. The ancestors of whales and dolphins were four-legged, land-dwelling mammals, “returning” to the aquatic environment roughly 54 million years ago (Gingerich et al. 1983; Gingerich et al. 1994; Bajpai and Gingerich 1998; Gingerich 2003; Uhen 2007b; Uhen 2007a; Uhen 2010). Concomitant with the loss of external hindlimbs and the evolution of streamlined fluke-powered swimming, the cetacean pelvic girdle evolved to a highly reduced state (Fig. 1A) over the span of roughly 7 million years (Gatesy et al. 2013; McGowen et al. 2014). Cetacean pelves no longer share a symphysis at the sacrum, no longer contact the vertebral column, have lost their acetabulum and obturator foramen, and evolved to such small size that there is still no general consensus as to which of the three fused bones found in their terrestrial ancestor (ischium, ilium and/or pubis) they derive from (Cuvier 1823; Abel 1907; Lönnberg 1911; Lönnberg 1938; Yablokov et al. 1974; Tajima et al. 2004).

Figure 1.

Figure 1

Open in a new tab

Skeletal anatomy of the bottlenose dolphin (Tursiops truncatus). (A) External hindlimbs have been lost, and the pelvic girdle is reduced to a pair of small bones that no longer articulates with the skeleton. An adult male of approximately 3 meters. Artwork: Carl Buell. Used with permission from John Gatesy. (B) A closeup of the pelvic bone (cream colored), showing the soft tissue anatomy (pink). Labels indicate one of the paired ischiocavernosus muscles, the retractor penis muscle (Retractor) that holds the penis inside the body, and the genital slit of a male dolphin. Despite recent reports, the retractor penis muscle does not attach to the pelvic bone, and its sole function appears to be for holding the penis inside the body (Ommanney 1932); a male erection is therefore accompanied by relaxation of the retractor penis muscle, allowing the penis to extend outside the body. (C) Laser scans from paired pelvic bones (dorsal view, anterior towards top of figure) with 962 semi-landmarks (green) (see Figures S3–S4 for more detail). These pelvic bones are approximately 11 cm long. (D) Laser scans and semi-landmarks from the anterior-most pair of ribs. These rib bones are approximately 14 cm long.

Due to their highly reduced state, cetacean pelvic bones are sometimes thought of as “useless vestiges” of their land-dwelling ancestry (Curtis and Barnes 1989). However, all 92 extant cetacean species except two (Kogia sima and K. breviceps, which appear to have replaced their pelvic bones with functional cartilaginous sheets, Benham 1901) have retained their pelvic bones. Furthermore, the anatomy of cetacean pelvic bones reveals important roles in male reproductive function. The paired pelvic bones anchor the genitalia and the paired ischiocavernosus muscles, which control the penis (Struthers 1881; Delage 1885; Abel 1907; Meek 1918; Anthony 1922; Ommanney 1932; Slijper 1966; Pabst et al. 1998; Tajima et al. 2004; Rommel et al. 2007; Thewissen et al. 2009) (Figure 1B). In male cetaceans, the paired ischiocavernosus muscles insert deeply towards the distal end of the penis, while proximally encapsulating the paired crus of the penis which anchor to each pelvic bone (Ommanney 1932; Arvy 1978; De Guise et al. 1994) (Figure 1B). The ischiocavernosus muscles appear to maneuver the penis by pulling it to one side or the other (Delage 1885) and may also maintain erection by compressing the corpus cavernosum proximally (Ommanney 1932). The ischiocavernosus muscles may work in conjunction with other soft tissue innovations to control the penis, but work in other systems suggests the ischiocavernosus muscles are fundamentally important in penis movements. In rats, surgically detaching the ischiocavernosus muscles results in severe penile dysfunction; males are unable to perform “flips” (rapid dorsiflexion of the penis) and as a result they cannot penetrate the vagina with their penis (Sachs 1982; Hart and Melese-D’Hospital 1983). In humans, anesthetizing the ischiocavernosus muscles results in a weaker erection (Claes et al. 1996).

The ischiocavernosus muscles attach to the pelvis in all mammals, but cetacean pelvic bones are unique because they are no longer constrained by sacral and hind limb attachments nor by hind limb locomotion, potentially freeing them to diverge according to penis morphology. Penis length and control may evolve to counteract mating avoidance behavior. In at least some cetacean species, females will surface “belly-up” to avoid mating with coercive males. Male cetaceans can apparently maneuver their penises to overcome this female mating resistance behavior (Mate et al. 2005), and since penis length varies among marine mammals (Brownell and Ralls 1986; Fitzpatrick et al. 2012) the supporting pelvic bones may be an important part of male anatomy. Furthermore, cetacean penises lack a baculum (Slijper 1966) (a bony os penis found in many mammals) as expected given its apparent flexibility.

Given evolutionary and anatomical evidence, we hypothesized that evolutionary patterns of cetacean pelvic bones have been driven by sexual selection, for instance via selection for larger penises, which would require larger ischiocavernosus muscles and their attachment sites. If sexual selection favors males that control their penises in novel ways, we also hypothesized that pelvic bone shape divergence (independent of size) would correlate to inferred mating system. We developed a novel morphometric pipeline to test these hypotheses using bones from 130 adult specimens taken from 29 species (Table S1), and report two main findings. First, cetaceans from relatively promiscuous mating systems (inferred by the well-established proxy of relative testes size, see below) have relatively long penises and relatively large pelvic bones, presumably to anchor relatively large ischiocavernosus muscles. Second, pelvic bone shape (independent of size) has diverged more in species pairs with larger inferred differences in mating system. Our study suggests that far from being mere relics of a terrestrial past, cetacean pelvic bones are targets of sexual selection.

Materials and Methods

Data and associated code from the entire pipeline described below are available on the Dryad Digital Repository (http://dx.doi.org/10.5061/dryad.[NNNN]).

Estimating the intensity of postcopulatory sexual selection across cetaceans

Direct data on mating system are lacking for most cetaceans. Therefore, we indirectly estimated the strength of post-copulatory sexual selection from testes mass data gathered from the literature. Across a broad taxonomic diversity, males of species with relatively intense postcopulatory sexual selection tend to possess relatively large testes compared to their body size, including mammals (Harcourt et al. 1981; Kenagy and Trombulak 1986; Møller 1989; Stockley and Purvis 1993; Ramm et al. 2005; Firman and Simmons 2008; Simmons and Fitzpatrick 2012), fish (Stockley et al. 1997), and insects (Gage 1994; Hosken 2001; Simmons and García-González 2008). Large testes presumably represent a costly though adaptive trait as males vie for fertilization under competitive conditions. This pattern has been confirmed within species, among populations that differ in their inferred mating system (Firman and Simmons 2008), and upon experimental exposure to variable mating regimes (Hosken and Ward 2001; Pitnick et al. 2001).

In the modern era, most cetacean specimens come from beached animals in various states of degradation, so neither testes mass nor body mass can be reliably measured. Furthermore, testes regress outside of the breeding season. Therefore, we gathered maximum testes mass and body length from the literature for each species (Table S2). Using the phylogeny of McGowen et al. (2009), we estimated the residuals of maximum testes mass regressed onto maximum body length using phylogenetic generalized least squares (PGLS) implemented via the GLS procedure in the R package NLME, with a correlation structure that accounted for phylogenetic relatedness (Pagel 1999), using the corPagel procedure in the R package APE (Paradis et al. 2004). Maximum body mass was used in the regression of testes mass because those data were derived from freshly dead specimens, many from the whaling industry. The phylogenetic residuals of testes mass (Figures 2, S1) were used as a proxy of the strength of postcopulatory sexual selection within each species.

Figure 2.

Figure 2

Open in a new tab

Phylogeny of cetacean taxa (McGowen et al. 2009) with known maximum testes size and body length. Numbers to the left of parentheses indicate residual testes mass (calculated as in Fig. S1). Numbers in parentheses indicate the number of individuals sampled for male pelvic, male rib, female pelvic, and female rib bones, respectively. Images indicate representative body morphology (left column, artwork from Carl Buell, used with permission from John Gatesy), paired pelvic bones (middle column), and paired rib bones (right column) from a subset of species. N/A=rib bones could not be sampled from all museum specimens.

Analyzing relative penis length

Using the same phylogenetically controlled methods just described, we regressed penis length onto body length, using the data from Table 2 of Brownell and Ralls (1986), who reported on 10 baleen whale species. We then regressed the residuals of penis length onto testes mass residuals calculated above (Figure S2).

Defining semi-landmarks on bones

From museum collections, we gathered pelvic bones from 97 sexually mature males (24 species) (Figure 2, Table S1). Where possible (87 sexually mature males from 20 species), we included the anterior-most pair of vertebral ribs as a negative control as they are not linked to genital musculature (Figure 2, Table S1). The first pair of ribs was chosen because it is a paired set of bones, and their distinctive shape makes them easy to identify from a specimen box containing the entire disarticulated skeleton. It should be noted that some cetacean species retain internal vestiges of femurs and/or tibiae (Struthers 1881; Howell 1970; Omura 1980) and in extremely rare cases individuals develop external hindlimbs (Andrews 1921; Sleptsov 1939; Ohsumi 1965; Berzin 1972). However, these additional bones do not attach to the ischiocavernosus muscles and are ignored here.

Cetacean pelvic bones are devoid of distinguishing landmarks (Figure 1), hampering traditional morphometric techniques. To overcome these challenges, we first scanned bones with a NextEngine HD Model 2020i 3-dimensional Laser Scanner (NextEngine Inc., Santa Monica, CA), which returns tens of thousands of x; y; z points from the surface of each bone. Using a variety of tools in computational geometry, we defined 962 semi-landmarks from digitized versions of the bones for downstream analyses (Figures 1C, 1D). This computational pipeline is graphically illustrated for pelvic bones (Figure S3) and ribs (Figure S4) separately.

It is important to consider the female side of sexual selection (Ah-King et al. 2014). As in males, female ischiocavernosus muscles originate on the pelvic bone, but insert instead on the clitoris. There are at least three predictions about how female pelvic bones should evolve. First, female pelvic bones may show no relationship to relative testes mass of their species, for example if female pelvic bones are nonfunctional. Second, even if female pelvic bones are non-functional, they may follow the patterns observed in males simply through shared developmental programs. Third, females may experience their own unique selective pressures associated with pelvic bones if they are functional. We tested these predictions by repeating the above analyses on female pelvic (33 sexually mature females from 17 species) and rib (27 females from 12 species) bones (Figure 2, Table S1).

The evolution of bone size

Bone size was estimated with centroid size, the square root of the sum of squared distances of the 962 semi-landmarks from their centroid. We quantified technical replication by randomly choosing 41 bones (21 pelvic bones, 20 ribs) to scan more than once. One bone was scanned 11 times, the rest scanned twice, each time removing the bone from the scanner and reloading it. The median coefficient of variation (unbiased standard deviation/mean) for centroid size was 0.0094 for pelvic bones, 0.0090 for ribs, indicating our methods of measuring size were highly repeatable.

Two different analyses were performed to test whether bone size evolved in a correlated manner with residual testes mass. First, we calculated the phylogenetic residuals of species-averaged centroid size regressed onto species-averaged body length, then tested whether those residuals were correlated to testes mass residuals, all using PGLS methodology already described. Unlike our regression of maximal testes mass onto maximal body mass above, here we regressed residual centroid size onto each specimen’s body length. Accurate body mass for museum specimens were rarely recorded, and would be downwardly biased due to inherent pathologies and decomposition of dead-stranded individuals, the source of most museum specimens.

The simple application of PGLS ignored several complexities in the data, including missing data (not all target bones are present in museum collections) and uneven sampling of species (Table S1). Furthermore, as a correlation between residual pelvic size and residual testes mass, the underlying model is not clear. Therefore, we developed a second, customized phylogenetic model to explicitly model correlated trait evolution (Harmon et al. 2008) in which log body length, testes size, and left and right pelvic and rib centroid sizes evolve as a multivariate Brownian motion on the cetacean phylogeny (McGowen et al. 2009), and individual samples were displaced by independent amounts from their species means. This is a standard model of correlated trait evolution (Revell and Collar 2009) except for explicit modeling of intraspecific variation and missing data. To do Bayesian inference, we put zero-mean Gaussian priors on the entries of the Cholesky decomposition of the covariance matrix and ran a Markov chain Monte Carlo sampler to estimate the posterior distribution of the parameters. The mathematical details and parameter estimates appear in S1 Materials and Methods.

The evolution of bone shape

We quantified shape difference between all possible pairs of pelvic bones, or between all possible pairs of rib bones in a Generalized Procrustes framework, which standardizes each set of 962 semi-landmarks to a common size, translates them to a common origin, then optimally rotates semi-landmark coordinates to minimize their Procrustes distance, which is the square root of the sum of the squared distances between corresponding landmarks (Rohlf and Bookstein 1990; Slice 2007). During Procrustes superimposition, our semi-landmarks were allowed to “slide” along the bones’ surfaces using the function GPAGEN in the R package GEOMORPH (Adams and Otárola-Castillo 2012). Sliding semi-landmarks is a well-established morphometric technique to improve alignment of corresponding anatomical regions lacking individual landmark homology (Bookstein 1996; Bookstein 1997; Gunz et al. 2005; Mitteroecker and Gunz 2009). In short, sliding semi-landmarks accompanies uncertainty in specific placement of landmarks. Left-sided bones were reflected prior to aligning.

Technical replication was estimated using the same re-scans described above. The median coefficient of variation for shape differences between each re-scan and all the other bones in the sample was 0.0194 for pelvic bones, 0.0342 for ribs, indicating our methods for measuring shape divergence were highly repeatable.

Sexual dimorphism

All analyses described above were performed for males and females separately. In addition, four different distance-based ANOVA’s (McArdle and Anderson 2001) were performed to test the contribution of sex to variance in pelvic or rib bones size or shape, using the function ADONIS in the R package VEGAN (Oksanen et al. 2013), with significance determined with 10,000 permutations.

Results

Estimating the intensity of postcopulatory sexual selection across cetaceans

All else equal, species with larger testes mass residuals are expected to experience a more promiscuous mating system. This prediction has been confirmed in multiple systems (see above), and aligns well with the scant amount of independent data on mating system in cetaceans. For example, Pontoporia blainvillei (the franciscana) has the smallest absolute and residual testes mass (Table S2), and is thought to be monogamous (Danilewicz et al. 2004). In contrast, direct behavioral and genetic observations suggests three species - Eubalaena glacialis (North Atlantic right whale) (Mate et al. 2005; Frasier et al. 2007; Frasier et al. 2013), Balaena mysticetus (bowhead whale) (Würsig and Clark 1993), and Lagenorhynchus obscurus (the dusky dolphin) (van Waerebeek and Read 1994) - are promiscuous, and all three have large absolute and residual testes (Table S2). We tested all subsequent predictions of size and shape evolution in the framework of residual testes mass, with the idea that relatively large residuals indicate species with relatively more promiscuous mating systems.

Relative penis length increases with relative testis mass

Species with large relative testes mass have significantly larger penises compared to their body length (phylogenetically controlled p<10−4, r=0.65, Figure S2). While the ultimate mechanism behind this correlation is not clear, one possibility is that males with longer penises can better overcome female resistance behavior in relatively promiscuous species, behaviors that were observed by Mate et al. (2005). Or perhaps female reproductive tracts are more convoluted in more promiscuous species, favoring males that can deposit sperm closer to the sites of fertilization. Longer genitalia in relation to sexual selection has been observed in other mammalian taxa (Miller and Burton 2001; Lüpold et al. 2004; Kinahan et al. 2007; Fitzpatrick et al. 2012). Whatever the underlying cause, we hypothesized that species with relatively large testes must have relatively large ischiocavernosus muscles to control their relatively large penises, which in turn require relatively large pelvic bones to serve as anchors. We note that the penis length data only derive from baleen whales. Any uncertainty about the relationship between residual testes mass and residual penis length in toothed whales will only introduce noise into our studies of correlated trait evolution across the full phylogeny, making our conclusions below conservative.

Relative size of pelvic bones increases with relative testis mass

Two different methods demonstrated that pelvic bone size increased along with testes mass. First, using PGLS methodology, relative pelvic size (log pelvic bone centroid size phylogenetically regressed onto log body length per specimen, averaged per species, Figure S5) was positively correlated with relative testes mass (phylogenetically controlled p=0.0005, r=0.60, Figure 3A), a pattern not observed in ribs (phylogenetically controlled p=0.98, r=0.04, Figures 3B, S6). A Q-Q plot identified three outlier species (Ziphius cavirostris, Feresa attenuata, Eschrichtius robustus) that contributed disproportionately to the regression (Fig. 3), but the PGLS was still significant after removing them (phylogenetically controlled p<10−4, r=0.75). The PGLS was also significant if we confined the analysis to those specimens for which both pelvic and rib data existed (phylogenetically controlled p=0.0048, r=0.58).

Figure 3.

Figure 3

Open in a new tab

Among sexually mature males, residual centroid size (species-average centroid size regressed onto body length) was positively correlated with residual testes mass (using PGLS methodology) in (A) pelvic bones, but not (B) ribs. Species: 1-Balaenoptera acutorostrata, 2-Balaenoptera musculus, 3-Delphinus capensis, 4-Delphinus delphis, 5-Eschrichtius robustus, 6-Eubalaena glacialis, 7-Feresa attenuata, 8-Grampus griseus, 9-Inia geoffrensis, 10-Lagenorhynchus acutus, 11-Lagenorhynchus obliquidens, 12-Lissodelphis borealis, 13-Phocoena phocoena, 14-Phocoenoides dalli, 15-Pontoporia blainvillei, 16-Pseudorca crassidens, 17-Stenella attenuata, 18-Stenella coeruleoalba, 19-Stenella frontalis, 20-Stenella longirostris, 21-Steno bredanensis, 22-Tursiops truncatus, 23-Ziphius cavirostris.

Second, our customized phylogenetic model demonstrated that after removing their common correlation to body length, male pelvic bone centroid sizes were positively correlated to testes mass (correlation rho=0.67, 95% credible interval: 0.25 to 0.90, Figure 4A, Tables S3–S4). In other words, species with relatively large testes have relatively large pelvic bones for their body length. There was no correlation between rib centroid size and testes size (correlation rho=0.07, 95% credible interval: −0.51 to 0.62, Figure 4B, Tables S3–S4). Results were qualitatively similar if we only analyzed those individuals represented in both the pelvic and the rib datasets (Figure S7) although we narrowly lose statistical significance at 95% for the correlation between pelvic bone centroid size and testes (correlation rho=0.58, 95% credible interval: −0.05 to 0.92, Tables S5–S6).

Figure 4.

Figure 4

Open in a new tab

The results of a customized Bayesian model of correlated evolution between traits. (A) All 1000 correlation coefficients sampled from the marginal posterior distributions showed that shifts in relative testes mass positively predicted shifts in pelvic centroid size (correlation rho=0.67, 95% credible interval: 0.25 to 0.90). (B) In contrast, shifts in testes size did not predict shifts in rib centroid size (correlation rho=0.05, 95% credible interval: −0.38 to 0.48).

To limit the ecological variation across the full phylogeny, and to specifically focus on species where intraspecific variation could be estimated, we re-analyzed two independent species pairs with relatively large sample sizes: Phocoena phocoena (N=6 sexually mature males) vs. Phocoenoides dalli (N=8) and Delphinus capensis (N=24) vs. D. delphinus (N=8). For both species pairs, residual pelvic bone centroid size was significantly larger in the species with the larger residual testes size (t=8.69, 3.40; df=10, 29; p<10−5, 0.002 for the two species pairs, respectively), a pattern not observed in ribs (t=0.88, 1.08; df=7, 26; p=0.41, 0.29, respectively). The two Delphinus species are so closely related that they were only recognized as separate species 20 years ago (Heyning and Perrin 1994); even in this species pair, the taxon with the larger residual teses size displayed larger residual pelvic bone size.

In sum, cetaceans with relatively large testes have relatively large pelvic bones, but not ribs. Relatively large pelvic bones provide more surface area for attachment of the ischiocavernosus muscles, offering one mechanism by which males could achieve enhanced maneuverability of relatively large penises.

Pelvic bone shape diverges more rapidly among species that have diverged in inferred mating system

Among 9 independent species pairs for which both pelvic bones and rib bones could be analyzed, pelvic bone shape divergence (independent of size) was positively correlated with divergence in inferred mating system (p=0.008, r=0.81, Fig. 5A). As with the analyses of size presented above, such a relationship did not hold for ribs (p=0.72, r=−0.14, Figure 5B), demonstrating a unique link between inferred mating system and pelvic bone shape evolution.

Figure 5.

Figure 5

Open in a new tab

(A) Among nine independent species pairs (filled circles each represent a species pair), pelvic bone shape divergence was positively correlated with divergence in testes residuals (p=0.008, r=0.81). (B) In contrast, rib shape divergence was not correlated with divergence in testes size residuals (p=0.72, r=−0.14). For completeness, the data for all pairwise contrasts, including non-independent contrasts, is shown in gray circles, although they were not included in tests of correlation. Only sexually mature males for which both pelvic and rib bones could be sampled were included; Species pairs: 1=Delphinus capensis vs. Delphinus delphis, 2=Feresa attenuata vs. Pseudorca crassidens, 3=Grampus griseus vs. Steno bredanensis, 4=Inia geoffrensis vs. Pontoporia blainvillei, 5=Lagenorhynchus acutus vs. Ziphius cavirostris, 6=Lagenorhynchus obliquidens vs. Lissodelphis borealis, 7=Phocoenoides dalli vs. Phocoena phocoena, 8=Stenella attenuata vs. Stenella longirostris, 9=Stenella coeruleoalba vs. Stenella frontalis.

Sexual dimorphism

Interestingly, female pelvic bone centroid size residuals were positively correlated with residual testes mass of their species, in both the PGLS and the customized phylogenetic models described (PGLS: p=0.002, r=0.68, Figure S8; customized model: rho=0.77, 95% credible interval: 0.30 to 0.99, Figure S9, Tables S7–S8) a pattern not observed in ribs (PGLS: p=0.90, r=−0.10, Figure S8; customized model: rho=0.20, 95% credible interval: −0.58 to 0.80, Figure S9, Tables S7–S8), just as observed in males. In other words, pelvic bones are relatively large in females of species that have males with relatively large testes.

The similarity of evolutionary patterns observed in males and females could be due to shared developmental programs. However, both pelvic bones and ribs were significantly sexually dimorphic in both size and shape (p<0.05 in all cases, Tables S9–S12), consistent with previous reports (van Bree 1973; Perrin 1975; Arvy 1979; Andersen et al. 1992). Taken together, these results suggest that pelvic bones function in unique aspects of male and female reproductive ecology. As discussed above, pelvic bones anchor ischiocavernosus muscles in females, which insert on the clitoris. It is possible that clitoral movements play a role in female choice, potentially placing the female pelvic bone in the arena of sexual selection. At the moment this hypothesis remains speculative.

Discussion

Under certain models, sexual selection favors females that make it difficult – behaviorally, morphologically, or otherwise - for males to fertilize their ova, which subsequently favors males that overcome these defenses (Eberhard 1985; Andersson 1994; Eberhard 2009). Such mechanisms may yield indirect benefits to females, through acquisition of good genes for her progeny and/or her ability to produce competitive sons. The strength of such coevolutionary dynamics is expected to intensify in relatively promiscuous species. For example, genital morphology evolves relatively rapidly in promiscuous vs. monogamous species (Arnqvist 1998). In the context of cetaceans, female avoidance behavior may induce selection on males to overcome such defenses. Because direct observations of cetacean mating behavior are scant, this hypothesis remains speculative. However, males have been observed to counteract female avoidance behavior through dexterous control of their penises (Mate et al. 2005).

Across the cetacean phylogeny, both size and shape of pelvic bones are evolutionary correlated to relative testes mass, a strong indication of the strength of post-copulatory sexual selection (Harcourt et al. 1981; Kenagy and Trombulak 1986; Møller 1989; Gage 1994; Stockley et al. 1997; Hosken 2001; Ramm et al. 2005; Firman and Simmons 2008). One unifying hypothesis is that cetaceans that experience strong sexual selection have evolved relatively large penises, which require relatively large muscles and pelvic bones to serve as anchor sites for genital control. Sexual selection also appears to favor divergence in shape, perhaps allowing males to maneuver their penises in novel ways. Importantly, our study rejects a common assumption (mostly among non-cetacean biologists) that cetacean pelvic bones are “useless vestiges” (Curtis and Barnes 1989), and instead suggest they are a critical component of male, and possibly female, reproductive fitness. To our knowledge, this is the first example of sexual selection affecting the internal infrastructure affecting genital movements.

Supplementary Material

Supp Material

Figure S1. The regression of maximal recorded testis size onto maximal recorded body mass. The regression was drawn using the GLS procedure in the R package NLME, with a correlation structure that accounted for phylogenetic relatedness (Pagel 1999), using the corPagel procedure in the R package APE (Paradis et al. 2004). The phylogenetic residuals were used as a measure of relative testes size.

Figure S2. Whale species with relatively large testes have relatively long penises. Two separate regressions, each drawn using the GLS procedure in the R package NLME and a correlation structure accounting for phylogenetic relatedness (Pagel 1999) using the corPagel procedure in the R package APE (Paradis et al. 2004), were drawn to derive the residuals of penis length onto body mass, and testes mass onto body mass. A third phylogenetic regression (pictured here) was used to test the correlation of residual penis length with residual testis mass. Raw data taken from Table 2 of (Brownell and Ralls 1986), and analyzed in the phylogenetic framework of (McGowen et al. 2009).

Figure S3. A series of tools in computational geometry were deployed to define landmarks from pelvic bone scans. First, the two furthest points apart were found to represent the most posterior and most anterior points (point 1 and 3). The entire point cloud was transformed so that point 1 was x=0, y=0, z=0, and point 3 was x=0, y=0, with some positive value of z. A new z-axis was drawn between point 1 and 3 (red line), and the point furthest from this line found (point 2), and the entire point cloud transformed so that point 2 was x=some positive value, y=0, z=some positive value. After this initial transformation, we sampled points that fell within the most posterior 0.5% of the z axis, the most anterior 0.5% of the z axis, or the middle 0.5% of the z axis (typically, several hundred points sampled each region), and calculated the centroid of their respective convex hulls. This second transformation accounted for variation in the placement of the first three points. Then, 60 evenly-spaced slices of points were sampled along the z axis, where each slice thickness was 0.5% the length of the z-axis. One slice appears in the right panel as an example. For each slice, the midpoint of the convex hull was calculated (green point in middle of right panel). Then, moving in increments of 22.5 degrees from the x-axis (only the first increment shown in detail), we determined the two points straddling the line extending from the centroid (two points shown in blue, one slightly obscured, with connecting line drawn in blue). A landmark was defined as the intersection of the black and blue lines (green point) and named according to its slice (i.e., P30) and its angle from the x-axis (i.e., 22.5). Only P30 22.5 is shown in detail: the other green points are shown for completeness. All green points correspond to the green points shown in Fig. 1C of the manuscript. Only the green points were included in downstream analyses; all other points were discarded.

Figure S4. A series of tools in computational geometry were deployed to define landmarks from rib bone scans. First, all points were initially transformed according to the same three initial points described in Fig. S3 for pelvic bones. Then, a fulcrum (black sphere) was placed 1/3 up the length of the z-axis, and the bone scan divided into 60 slices of equal angle from that fulcrum. For each slice (the 30th slice shown as an example in right panel), the centroid of the convex hull of the 20% most medial points (leftmost blue sphere) and 20% most lateral points (rightmost blue sphere) points was used to draw a line. We identified the point on that line which was closest to a point in the original scan (rightmost/leftmost green spheres, right panel). This line was then divided evenly (black spheres) and lines perpendicular to the black spheres computed. All points from the original scan within a certain distance to the perpendicular lines were found, projected onto the perpendicular line and the midpoint computed (orange spheres on perpendicular lines, right panel). The orange sphere was projected onto the plane parallel to the figure to find a third point with which to draw a plane (indicated by orange triangles, right panel). Those orange triangles allowed us to divide the bone into anterior and posterior halves and to account for complex curvatures. The corresponding green spheres on the perpendicular lines were determined as done for the medial and lateral green spheres. All green spheres are named as described in Fig. S3, and correspond to the green spheres shown in Fig. 1D of the manuscript. Only the green spheres were included in downstream analyses; all other points were discarded.

Figure S5. Centroid size regressed onto max body length. For each species (sexually mature males only), an average centroid size was calculated, then regressed onto body length using the gls procedure in the R package nlme, with a correlation structure that accounted for phylogenetic relatedness (44), using the corPagel procedure in the R package ape (45). The phylogenetic residuals of centroid size were used as a measure of relative pelvic size.

Figure S6. For each species (sexually mature males only), an average rib centroid size was calculated, then regressed onto body length using the gls procedure in the R package nlme, with a correlation structure that accounted for phylogenetic relatedness (Pagel 1999), using the corPagel procedure in the R package ape (Paradis et al. 2004). The phylogenetic residuals of centroid size were used as a measure of relative rib size.

Figure S7. The marginal posterior distributions of correlation coefficients do not change when restricting to bones from adult male cetaceans for which we have both data for both ribs and pelvic bones, which excludes baleen whales. (compare to figure 4 of the main text).

Figure S8. As in males (see Fig. 3 of manuscript), among sexually mature females, residual centroid size (species-average centroid size regressed onto body length) was positively correlated with residual testes mass of males from their species (maximum species testes mass regressed onto maximum body length) in (A) pelvic bones, but not (B) ribs. Species: 1–Balaenoptera acutorostrata, 2–Balaenoptera musculus, 3–Balaenoptera physalus, 4–Delphinus capensis, 5–Delphinus delphis, 6–Eubalaena glacialis, 7–Globicephala melas, 8–Lagenorhynchus acutus, 9–Lagenorhynchus obliquidens, 10–Mesoplodon carlhubbsi, 11–Peponocephala electra, 12–Phocoena phocoena, 13–Phocoenoides dalli, 14–Pontoporia blainvillei, 15–Stenella attenuata, 16–Tursiops truncatus, 17–Ziphius cavirostris.

Figure S9. As observed for males (figure 4 of manuscript), the marginal posterior distributions of correlations between changes in female pelvic bone size was significantly correlated with shifts towards larger testis size. Under a phylogenetic model of correlated trait evolution (supplementary methods), the size of the pelvic bone is positively correlated with shifts towards larger testes after accounting for body size evolution (red), while rib bone size does not show this correlation (gray). Only sexually mature females were included in this analysis.

Table S1. Individual level data from bone scans. Only sexually mature individuals included here and in all analyses. NA indicates samples that were absent from museum collection or could not be scanned. Numbers presented for bones are centroid sizes. Museum source indicated in specimen column (BMNH=British Museum of Natural History, CCSN=Cape Cod Stranding Network, LACM=Los Angeles County Natural History Museum, MAL=Marine Animal Life, MH=New England Aquarium, MJM=Michael J. Moore, UMA=University of Massachusetts Amherst, USNM=United States Natural History Museum (Smithsonian), UWBM=University of Washington Burke Museum). Specimens in bold were scanned multiple times to assess technical replication (one juvenile not shown).

Table S2. Morphological data gathered from literature for sexually mature males.

Table S3. Posterior means and quantiles of the parameters of the model presented in equations (7) and (8) of supplemental methods, estimated using only bones from adult males.

Table S4. Marginal posterior distributions of correlations, with length fixed, between changes in rib size, pelvic bone size, and testes size, estimated using only bones from adult males.

Table S5. Posterior means and quantiles of the parameters given the dataset consisting only of bones from males for which we have both ribs and pelvic bones.

Table S6. Marginal posterior distributions of correlations, with length fixed, between changes in rib size, pelvic bone size, and testes size, given only data for bones in males for which we have both ribs and pelvic bones.

Table S7. Posterior means and quantiles of the parameters, for bones from females only.

Table S8. Correlations from females: Marginal posterior distributions of correlations, with length fixed, between changes in rib size, pelvic bone size, and testes size, given bones from females only.

Table S9. Distance-based ANOVA of pairwise differences in pelvic bone centroid size. Results from a nested ANOVA performed on the pairwise distance matrix of relative centroid size (centroid size divided by body length). Significance determined with 10,000 permutations, implemented as “adonis (distance matrix ~ species + sex within species + specimen within sex within species)” in the ADONIS function of the Oksanen et al. (Oksanen et al. 2013) package in R.

Table S10. Distance-based ANOVA of pairwise differences in rib bone centroid size. Analysis as described in Table S9.

Table S11. Distance-based ANOVA of pairwise differences in pelvic bone shape. Analysis as described in Table S9.

Table S12. Distance-based ANOVA of pairwise differences in rib bone shape. Analysis as described in Table S9.

Acknowledgments

For discussions, we thank the USC biodiversity group, John Fitzpatrick, Jeffrey Good, and Bret Payseur. John Gatesy granted permission to reprint cetacean illustrations. For assistance in acquiring specimens, we thank Charley Potter (Smithsonian Institution), Tom French (Massachusetts Division of Fisheries and Wildlife), Jeff Bradley (Burke Museum), Richard Sabin (British Museum), Kate Doyle (University of Massachusetts, Amherst) and Dave Janiger (LA County Museum of Natural History). Michael McGowen provided phylogenetic data. USC’s Center for High-Performance Computing and Communications provided computational resources. This work was supported by USC startup funds (MDD), NIH grant #1R01GM098536 (MDD) and the William Cheney, Jr. Memorial Fund for Mammalogy (JPD). All original data and analytical code have been deposited in the Dryad Digital Repository (http://dx.doi.org/10.5061/dryad.[NNNN]).

Footnotes

Data and associated code from the entire pipeline described below are available on the Dryad Digital Repository (http://dx.doi.org/10.5061/dryad.[NNNN]).

Supplemental Information

Supplemental Information includes nine figures, twelve tables, and Supplemental Experimental Procedures and can be found with this article online at http:NNNN.

References

  1. Abel O. Die Morphologie der Hüftbeinrudimente der Cetaceen. Denkschriften der Kaiserlichen Akadamie der Wissenschaften. 1907;81:139–195. [Google Scholar]
  2. Adams DC, Otárola-Castillo E. R package version 1.0. The Comprehensive R Network (CRAN); 2012. Package ‘geomorph’: Geometric morphometric analysis of 2d/3d landmark data. [Google Scholar]
  3. Ah-King M, Barron AB, Herberstein ME. Genital Evolution: Why Are Females Still Understudied? PLoS Biol. 2014;12:e1001851. doi: 10.1371/journal.pbio.1001851. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Andersen D, Kinze CC, Skov J. The use of pelvic bones in the harbour porpoise Phocoena phocoena as an indication of sexual maturity. Lutra. 1992;35:105–112. [Google Scholar]
  5. Andersson M. Sexual selection. Princeton University Press; Princeton, New Jersey: 1994. [Google Scholar]
  6. Andrews RC. A remarkable case of external hind limbs in a humpback whale. American Museum Novitates. 1921;9:1–6. [Google Scholar]
  7. Anthony R. Recherches anatomiques sur l’appareil genito-urinaire male du Mesoplodon et des cetaces en general. Memorias del Instituto Espanol de Oceanografia Madrid. 1922;3:35–216. [Google Scholar]
  8. Arnqvist G. Comparative evidence for the evolution of genitalia by sexual selection. Nature. 1998;393:784–786. [Google Scholar]
  9. Arnqvist G, Rowe L. Sexual conflict. Princeton University Press; Princeton, New Jersey: 2005. [Google Scholar]
  10. Arvy L. Le pénis de Cétacés. Mammalia. 1978;42:491–509. [Google Scholar]
  11. Arvy L. The abdominal bones of cetaceans. Investigations on Cetacea. 1979;10:215–227. [Google Scholar]
  12. Bajpai S, Gingerich PD. A new Eocene archaeocete (Mammalia, Cetacea) from India and the time of origin of whales. Proceedings of the National Academy of Sciences. 1998;95:15464–15468. doi: 10.1073/pnas.95.26.15464. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Baumgardner DJ, Hartung TG, Sawrey DK, Webster DG, Dewsbury DA. Muroid copulatory plugs and female reproductive tracts: a comparative investigation. J Mammal. 1982;63:110–117. [Google Scholar]
  14. Benham WB. On the anatomy of Cogia breviceps. Proceedings of the Zoological Society of London. 1901;71:107–134. [Google Scholar]
  15. Berzin AA. Israel Program for Scientific Translations, Jerusalem (Translated from Russian) 1972. The Sperm Whale (Kashalot) [Google Scholar]
  16. Bookstein FJ. Landmark methods for forms without landmarks: morphometrics of group differences in outline shape. Medical Image Analysis. 1997;1:225–243. doi: 10.1016/s1361-8415(97)85012-8. [DOI] [PubMed] [Google Scholar]
  17. Bookstein FL. Biometrics, biomathematics and the morphometric synthesis. Bulletin of Mathematical Biology. 1996;58:313–365. doi: 10.1007/BF02458311. [DOI] [PubMed] [Google Scholar]
  18. Brownell RL, Jr, Ralls K. Potential for sperm competition in baleen whales. Rep Int Whal Commn. 1986;8:97–112. [Google Scholar]
  19. Claes H, Bijnens B, Baert L. The hemodynamic influence of the ischiocavernosus muscles on erectile function. The Journal of urology. 1996;156:986–990. [PubMed] [Google Scholar]
  20. Curtis H, Barnes NS. Biology. 5. Worth Publishers; New York: 1989. [Google Scholar]
  21. Cuvier G. Chez G Dufour et E d’Ocagne. 1823. Recherches sur les ossemens fossiles: òu l’on rétablit les caractéres de plusieurs animaux dont les révolutions du globe ont détruit les espèces. [Google Scholar]
  22. Danilewicz D, Claver J, Pérez Carrera A, Secchi E, Fontoura N. Reproductive biology of male franciscanas (Pontoporia blainvillei) (Mammalia: Cetacea) from Rio Grande do Sul, southern Brazil. Fishery Bulletin. 2004;102:581–592. [Google Scholar]
  23. De Guise S, Bisaillon A, Seguin B, Lagace A. The anatomy of the male genital system of the beluga whale, Delphinapterus leucas, with special reference to the penis. Anatomia, histologia, embryologia. 1994;23:207–216. doi: 10.1111/j.1439-0264.1994.tb00469.x. [DOI] [PubMed] [Google Scholar]
  24. Delage Y. Histoire du Balaenoptera musculus échoué sur la Plage de Lagrune. Archives de Zoologie Expérimentale et Générale (2me Série) 1885;3:1–152. [Google Scholar]
  25. Dixson AF. Primate sexuality. Oxford University Press; New York: 1998. [Google Scholar]
  26. Eberhard W. Evolution of genitalia: theories, evidence, and new directions. Genetica. 2009;138:5–18. doi: 10.1007/s10709-009-9358-y. [DOI] [PubMed] [Google Scholar]
  27. Eberhard WG. Sexual selection and animal genitalia. Harvard University Press; Cambridge, Mass: 1985. [Google Scholar]
  28. Eberhard WG. Female control: sexual selection by cryptic female choice. Princeton University Press; Princeton, New Jersey: 1996. [Google Scholar]
  29. Firman RC, Simmons LW. The frequency of multiple paternity predicts variation in testes size among island populations of house mice. J Evol Biol. 2008;21:1524–1533. doi: 10.1111/j.1420-9101.2008.01612.x. [DOI] [PubMed] [Google Scholar]
  30. Fitzpatrick JL, Almbro M, Gonzalez-Voyer A, Kolm N, Simmons LW. Male contest competition and the coevolution of weaponry and testes in pinnipeds. Evolution. 2012;66:3595–3604. doi: 10.1111/j.1558-5646.2012.01713.x. [DOI] [PubMed] [Google Scholar]
  31. Frasier TR, Gillett RM, Hamilton PK, Brown MW, Kraus SD, White BN. Postcopulatory selection for dissimilar gametes maintains heterozygosity in the endangered North Atlantic right whale. Ecology and Evolution. 2013;3:3483–3494. doi: 10.1002/ece3.738. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Frasier TR, Hamilton PK, Brown MW, Conger LA, Knowlton AR, Marx MK, Slay CK, Kraus SD, White BN. Patterns of male reproductive success in a highly promiscuous whale species: the endangered North Atlantic right whale. Molecular Ecology. 2007;16:5277–5293. doi: 10.1111/j.1365-294X.2007.03570.x. [DOI] [PubMed] [Google Scholar]
  33. Gage MJG. Associations between body size, mating pattern, testis size and sperm lengths across butterflies. Proceedings of the Royal Society of London. Series B: Biological Sciences. 1994;258:247–254. [Google Scholar]
  34. Gatesy J, Geisler JH, Chang J, Buell C, Berta A, Meredith RW, Springer MS, McGowen MR. A phylogenetic blueprint for a modern whale. Mol Phylogenet Evol. 2013;66:479–506. doi: 10.1016/j.ympev.2012.10.012. [DOI] [PubMed] [Google Scholar]
  35. Gingerich PD. Land-to-sea transition in early whales: evolution of Eocene Archaeoceti (Cetacea) in relation to skeletal proportions and locomotion of living semiaquatic mammals. Paleobiology. 2003;29:429–454. [Google Scholar]
  36. Gingerich PD, Raza SM, Arif M, Anwar M, Zhou X. New whale from the Eocene of Pakistan and the origin of cetacean swimming. Nature. 1994;368:844–847. [Google Scholar]
  37. Gingerich PD, Wells NA, Russell DE, Shah SI. Origin of whales in epicontinental remnant seas: new evidence from the early Eocene of Pakistan. Science. 1983;220:403–406. doi: 10.1126/science.220.4595.403. [DOI] [PubMed] [Google Scholar]
  38. Gunz P, Mitteroecker P, Bookstein F. Semilandmarks in three dimensions. In: Slice DE, editor. Modern morphometrics in physical anthropology. 2005. pp. 73–98. [Google Scholar]
  39. Harcourt AH, Harvey PH, Larson SG, Short RV. Testis weight, body weight and breeding system in primates. Nature. 1981;293:55–57. doi: 10.1038/293055a0. [DOI] [PubMed] [Google Scholar]
  40. Harmon LJ, Weir JT, Brock CD, Glor RE, Challenger W. GEIGER: investigating evolutionary radiations. Bioinformatics (Oxford, England) 2008;24:129–131. doi: 10.1093/bioinformatics/btm538. [DOI] [PubMed] [Google Scholar]
  41. Hart BL, Melese-D’Hospital PY. Penile mechanisms and the role of the striated penile muscles in penile reflexes. Physiology & behavior. 1983;31:807–813. doi: 10.1016/0031-9384(83)90277-9. [DOI] [PubMed] [Google Scholar]
  42. Heyning JE, Perrin WF. Evidence for two species of common dolphins (genus Delphinus) from the eastern North Pacific. Contributions in Science, Natural History Museum of Los Angeles County. 1994;442:1–35. [Google Scholar]
  43. Higginson DM, Miller KB, Segraves KA, Pitnick S. Female reproductive tract form drives the evolution of complex sperm morphology. Proceedings of the National Academy of Sciences. 2012;109:4538–4543. doi: 10.1073/pnas.1111474109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Hosken D, Ward P. Experimental evidence for testis size evolution via sperm competition. Ecol Lett. 2001;4:10–13. [Google Scholar]
  45. Hosken DJ. Experimental evidence for testis size evolution via sperm competition. Ecol Lett. 2001;4:10–13. [Google Scholar]
  46. Hosken DJ, Stockley P. Sexual selection and genital evolution. Trends Ecol Evol. 2004;19:87–93. doi: 10.1016/j.tree.2003.11.012. [DOI] [PubMed] [Google Scholar]
  47. House CM, Simmons LW. Genital morphology and fertilization success in the dung beetle Onthophagus taurus: an example of sexually selected male genitalia. Proc R Soc B. 2003;270:447–455. doi: 10.1098/rspb.2002.2266. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. House CM, Simmons LW. The evolution of male genitalia: patterns of genetic variation and covariation in the genital sclerites of the dung beetle Onthophagus taurus. J Evol Biol. 2005;18:1281–1292. doi: 10.1111/j.1420-9101.2005.00926.x. [DOI] [PubMed] [Google Scholar]
  49. Howell AB. Aquatic mammals: their adaptation to life in the water. Dover Publications; New York, New York: 1970. [Google Scholar]
  50. Kenagy GJ, Trombulak SC. Size and function of mammalian testes in relation to body size. Journal of Mammalogy. 1986;67:1–22. [Google Scholar]
  51. Kinahan A, Bennett N, O’Riain M, Hart L, Bateman P. Size matters: genital allometry in an African mole-rat (Family: Bathyergidae) Evolutionary Ecology. 2007;21:201–213. [Google Scholar]
  52. Lönnberg E. The pelvic bones of some Cetacea. Arkw, for Zoo. 1911;7:8–15. [Google Scholar]
  53. Lönnberg E. Notes on the skeleton of Prodelphinus graffmani. Arkiv för Zoologi. 1938;30:1–21. [Google Scholar]
  54. Lüpold S, McElligott AG, Hosken DJ. Bat genitalia: allometry, variation and good genes. Biol J Linn Soc. 2004;83:497–507. [Google Scholar]
  55. Mate B, Duley P, Lagerquist B, Wenzel F, Stimpert A, Clapham P. Observations of a female North Atlantic right whale (Eubalaena glacialis) in simultaneous copulation with two males: supporting evidence for sperm competition. Aquatic Mammals. 2005;31:157–160. [Google Scholar]
  56. McArdle BH, Anderson MJ. Fitting multivariate models to community data: a comment on distance-based redundancy analysis. Ecology. 2001;82:290–297. [Google Scholar]
  57. McGowen MR, Gatesy J, Wildman DE. Molecular evolution tracks macroevolutionary transitions in Cetacea. TREE. 2014;29:336–346. doi: 10.1016/j.tree.2014.04.001. [DOI] [PubMed] [Google Scholar]
  58. McGowen MR, Spaulding M, Gatesy J. Divergence date estimation and a comprehensive molecular tree of extant cetaceans. Mol Phylogenet Evol. 2009;53:891–906. doi: 10.1016/j.ympev.2009.08.018. [DOI] [PubMed] [Google Scholar]
  59. Meek A. The reproductive organs of cetacea. Journal of anatomy. 1918;52:186–210. [PMC free article] [PubMed] [Google Scholar]
  60. Miller EH. Genitalic traits of mammals. In: Leonard JL, Córdoba-Aguilar A, editors. The evolution of primary sexual characters in animals. Oxford University Press; Oxford, UK: 2010. pp. 471–493. [Google Scholar]
  61. Miller EH, Burton LE. It’s all relative: allometry and variation in the baculum (os penis) of the harp seal, Pagophilus groenlandicus (Carnivora: Phocidae) Biol J Linn Soc. 2001;72:345–355. [Google Scholar]
  62. Mitteroecker P, Gunz P. Advances in geometric morphometrics. Evolutionary Biology. 2009;36:235–247. [Google Scholar]
  63. Møller AP. Ejaculate quality, testes size and sperm production in mammals. Functional Ecology. 1989;3:91–96. [Google Scholar]
  64. Ohsumi S. A dolphin (Stenella coeruleoalba) with protruded rudimentary hind limbs. Scientific Reports of the Whales Research Institute. 1965;19:135–136. [Google Scholar]
  65. Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, Simpson GL, Solymos P, Stevens MHH, Wagner H. R package version 2.0-7. The Comprehensive R Network (CRAN); 2013. Package ‘vegan’: Community Ecology. [Google Scholar]
  66. Ommanney FD. The urino-genital system in the fin whale. Discovery Reports. 1932;5:363–466. [Google Scholar]
  67. Omura H. Morphological study of pelvic bones of the minke whale from the Antarctic. Scientific Reports of the Whales Research Institute Tokyo. 1980;32:25–37. [Google Scholar]
  68. Pabst D, Rommel SA, McLellan WA. Evolution of thermoregulatory function in cetacean reproductive systems. In: Thewissen JGM, editor. The Emergence of Whales: Evolutionary Patterns in the Origin of Cetacea. Plenum; New York: 1998. pp. 379–390. [Google Scholar]
  69. Pagel M. Inferring the historical patterns of biological evolution. Nature. 1999;401:877–884. doi: 10.1038/44766. [DOI] [PubMed] [Google Scholar]
  70. Paradis E, Claude J, Strimmer K. APE: Analyses of Phylogenetics and Evolution in R language. Bioinformatics (Oxford, England) 2004;20:289–290. doi: 10.1093/bioinformatics/btg412. [DOI] [PubMed] [Google Scholar]
  71. Perrin WF. Variation of spotted and spinner porpoise (genus Stenella) in the Eastern Tropical Pacific and Hawaii. University of California Press; 1975. [Google Scholar]
  72. Pitnick S, Miller GT, Reagan J, Holland B. Males’ evolutionary responses to experimental removal of sexual selection. Proceedings of the Royal Society of London. Series B: Biological Sciences. 2001;268:1071–1080. doi: 10.1098/rspb.2001.1621. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Ramm SA. Sexual selection and genital evolution in mammals: a phylogenetic analysis of baculum length. The American Naturalist. 2007;169:360–369. doi: 10.1086/510688. [DOI] [PubMed] [Google Scholar]
  74. Ramm SA, Parker GA, Stockley P. Sperm competition and the evolution of male reproductive anatomy in rodents. Proc R Soc B. 2005;272:949–955. doi: 10.1098/rspb.2004.3048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Revell LJ, Collar DC. Phylogenetic analysis of the evolutionary correlation using likelihood. Evolution. 2009;63:1090–1100. doi: 10.1111/j.1558-5646.2009.00616.x. [DOI] [PubMed] [Google Scholar]
  76. Rohlf FJ, Bookstein FL. Proceedings of the Michigan Morphometrics Workshop. University of Michigan Museum of Zoology; Ann Arbor: 1990. [Google Scholar]
  77. Rommel SA, Pabst DA, McLellan WA. Functional anatomy of the cetacean reproductive system, with comparisons to the domestic dog. In: Miller DL, editor. Reproductive Biology and Phylogeny of Cetacea. Science Publishers; Enfield, NH: 2007. pp. 127–145. [Google Scholar]
  78. Sachs BD. Role of striated penile muscles in penile reflexes, copulation, and induction of pregnancy in the rat. J Reprod Fertil. 1982;66:433–443. doi: 10.1530/jrf.0.0660433. [DOI] [PubMed] [Google Scholar]
  79. Simmons LW. Sperm competition and its evolutionary consequences in the insects. Princeton University Press; Princeton, New Jersey: 2001. [Google Scholar]
  80. Simmons LW, Firman RC. Experimental evidence for the evolution of the mammalian baculum by sexual selection. Evolution. 2014;68:276–283. doi: 10.1111/evo.12229. [DOI] [PubMed] [Google Scholar]
  81. Simmons LW, Fitzpatrick JL. Sperm wars and the evolution of male fertility. Reprod. 2012;144:519–534. doi: 10.1530/REP-12-0285. [DOI] [PubMed] [Google Scholar]
  82. Simmons LW, García-González F. Evolutionary reduction in testes size and competitive fertilization success in response to the experimental removal of sexual selection in dung beetles. Evolution. 2008;62:2580–2591. doi: 10.1111/j.1558-5646.2008.00479.x. [DOI] [PubMed] [Google Scholar]
  83. Sleptsov MM. Rudimentary posterior fins in the Black Sea dolphin. Zoologicheskii Zhurnal. 1939;18:351–366. [Google Scholar]
  84. Slice DE. Geometrics morphometrics. Annual Review of Anthropology. 2007;36:261–281. [Google Scholar]
  85. Slijper EJ. Functional morphology of the reproductive system in Cetacea. In: Norris KS, editor. Whales, Dolphins and Porpoises. University of California Press; Los Angeles, CA: 1966. pp. 278–319. [Google Scholar]
  86. Stockley P, Gage MJG, Parker GA, Moller AP. Sperm competition in fishes: the evolution of testis size and ejaculate characteristics. The American Naturalist. 1997;149:933–954. doi: 10.1086/286031. [DOI] [PubMed] [Google Scholar]
  87. Stockley P, Purvis A. Sperm competition in mammals: a comparative study of male roles and relative investment in sperm production. Functional Ecology. 1993;7:560–570. [Google Scholar]
  88. Stockley P, Ramm SA, Sherborne AL, Thom MD, Paterson S, Hurst JL. Baculum morphology predicts reproductive success of male house mice under sexual selection. BMC biology. 2013;11:66. doi: 10.1186/1741-7007-11-66. [DOI] [PMC free article] [PubMed] [Google Scholar]
  89. Struthers J. On the bones, articulations, and muscles of the rudimentary hind-limb of the Greenland Right-Whale (Balaena mysticetus) Journal of Anatomy and Physiology. 1881;15:141–176. 302–321. [PMC free article] [PubMed] [Google Scholar]
  90. Tajima Y, Hayashi Y, Yamada TK. Comparative anatomical study on the relationships between the vestigial pelvic bones and the surrounding structures of finless porpoises (Neophocaena phocaenoides) The Journal of Veterinary Medical Science/the Japanese Society of Veterinary Science. 2004;66:761–766. doi: 10.1292/jvms.66.761. [DOI] [PubMed] [Google Scholar]
  91. Thewissen JGM, Cooper LN, George JC, Bajpai S. From land to water: the origin of whales, dolphins, and porpoises. Evo Edu Outreach. 2009;2:272–288. [Google Scholar]
  92. Uhen MD. Evolution of marine mammals: back to the sea after 300 million years. Anatomical Record. 2007a;290:514–522. doi: 10.1002/ar.20545. [DOI] [PubMed] [Google Scholar]
  93. Uhen MD. The terrestrial to aquatic transition in Cetacea. In: Anderson JS, Sues H-D, editors. Major Transitions in Vertebrate Evolution. Indiana University Press; Bloomington: 2007b. pp. 392–408. [Google Scholar]
  94. Uhen MD. The origin(s) of whales. Annual Review of Earth and Planetary Science. 2010;38:189–219. [Google Scholar]
  95. van Bree PJH. On the length and weight of the pelvic bones in the harbour porpoise, Phocoena phocoena (Linnaeus, 1758) in relation to sex and size. Lutra. 1973;15:8–12. [Google Scholar]
  96. van Waerebeek K, Read AJ. Reproduction of dusky dolphins, Lagenorhynchus obscurus, from coastal Peru. Journal of Mammalogy. 1994;75:1054–1062. [Google Scholar]
  97. Würsig B, Clark C. Behavior. In: Burns JJ, Montague JJ, Cowles CJ, editors. The Bowhead Whale. Society for Marine Mammalogy; 1993. pp. 157–199. [Google Scholar]
  98. Yablokov AV, Bel’kovich VM, Borisov VI. Whales and dolphins, parts 1 and 2. Joint Publications Research Service; 1974. JPRS-62150-1 and 2, NTIS, Translation from Russian. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp Material

Figure S1. The regression of maximal recorded testis size onto maximal recorded body mass. The regression was drawn using the GLS procedure in the R package NLME, with a correlation structure that accounted for phylogenetic relatedness (Pagel 1999), using the corPagel procedure in the R package APE (Paradis et al. 2004). The phylogenetic residuals were used as a measure of relative testes size.

Figure S2. Whale species with relatively large testes have relatively long penises. Two separate regressions, each drawn using the GLS procedure in the R package NLME and a correlation structure accounting for phylogenetic relatedness (Pagel 1999) using the corPagel procedure in the R package APE (Paradis et al. 2004), were drawn to derive the residuals of penis length onto body mass, and testes mass onto body mass. A third phylogenetic regression (pictured here) was used to test the correlation of residual penis length with residual testis mass. Raw data taken from Table 2 of (Brownell and Ralls 1986), and analyzed in the phylogenetic framework of (McGowen et al. 2009).

Figure S3. A series of tools in computational geometry were deployed to define landmarks from pelvic bone scans. First, the two furthest points apart were found to represent the most posterior and most anterior points (point 1 and 3). The entire point cloud was transformed so that point 1 was x=0, y=0, z=0, and point 3 was x=0, y=0, with some positive value of z. A new z-axis was drawn between point 1 and 3 (red line), and the point furthest from this line found (point 2), and the entire point cloud transformed so that point 2 was x=some positive value, y=0, z=some positive value. After this initial transformation, we sampled points that fell within the most posterior 0.5% of the z axis, the most anterior 0.5% of the z axis, or the middle 0.5% of the z axis (typically, several hundred points sampled each region), and calculated the centroid of their respective convex hulls. This second transformation accounted for variation in the placement of the first three points. Then, 60 evenly-spaced slices of points were sampled along the z axis, where each slice thickness was 0.5% the length of the z-axis. One slice appears in the right panel as an example. For each slice, the midpoint of the convex hull was calculated (green point in middle of right panel). Then, moving in increments of 22.5 degrees from the x-axis (only the first increment shown in detail), we determined the two points straddling the line extending from the centroid (two points shown in blue, one slightly obscured, with connecting line drawn in blue). A landmark was defined as the intersection of the black and blue lines (green point) and named according to its slice (i.e., P30) and its angle from the x-axis (i.e., 22.5). Only P30 22.5 is shown in detail: the other green points are shown for completeness. All green points correspond to the green points shown in Fig. 1C of the manuscript. Only the green points were included in downstream analyses; all other points were discarded.

Figure S4. A series of tools in computational geometry were deployed to define landmarks from rib bone scans. First, all points were initially transformed according to the same three initial points described in Fig. S3 for pelvic bones. Then, a fulcrum (black sphere) was placed 1/3 up the length of the z-axis, and the bone scan divided into 60 slices of equal angle from that fulcrum. For each slice (the 30th slice shown as an example in right panel), the centroid of the convex hull of the 20% most medial points (leftmost blue sphere) and 20% most lateral points (rightmost blue sphere) points was used to draw a line. We identified the point on that line which was closest to a point in the original scan (rightmost/leftmost green spheres, right panel). This line was then divided evenly (black spheres) and lines perpendicular to the black spheres computed. All points from the original scan within a certain distance to the perpendicular lines were found, projected onto the perpendicular line and the midpoint computed (orange spheres on perpendicular lines, right panel). The orange sphere was projected onto the plane parallel to the figure to find a third point with which to draw a plane (indicated by orange triangles, right panel). Those orange triangles allowed us to divide the bone into anterior and posterior halves and to account for complex curvatures. The corresponding green spheres on the perpendicular lines were determined as done for the medial and lateral green spheres. All green spheres are named as described in Fig. S3, and correspond to the green spheres shown in Fig. 1D of the manuscript. Only the green spheres were included in downstream analyses; all other points were discarded.

Figure S5. Centroid size regressed onto max body length. For each species (sexually mature males only), an average centroid size was calculated, then regressed onto body length using the gls procedure in the R package nlme, with a correlation structure that accounted for phylogenetic relatedness (44), using the corPagel procedure in the R package ape (45). The phylogenetic residuals of centroid size were used as a measure of relative pelvic size.

Figure S6. For each species (sexually mature males only), an average rib centroid size was calculated, then regressed onto body length using the gls procedure in the R package nlme, with a correlation structure that accounted for phylogenetic relatedness (Pagel 1999), using the corPagel procedure in the R package ape (Paradis et al. 2004). The phylogenetic residuals of centroid size were used as a measure of relative rib size.

Figure S7. The marginal posterior distributions of correlation coefficients do not change when restricting to bones from adult male cetaceans for which we have both data for both ribs and pelvic bones, which excludes baleen whales. (compare to figure 4 of the main text).

Figure S8. As in males (see Fig. 3 of manuscript), among sexually mature females, residual centroid size (species-average centroid size regressed onto body length) was positively correlated with residual testes mass of males from their species (maximum species testes mass regressed onto maximum body length) in (A) pelvic bones, but not (B) ribs. Species: 1–Balaenoptera acutorostrata, 2–Balaenoptera musculus, 3–Balaenoptera physalus, 4–Delphinus capensis, 5–Delphinus delphis, 6–Eubalaena glacialis, 7–Globicephala melas, 8–Lagenorhynchus acutus, 9–Lagenorhynchus obliquidens, 10–Mesoplodon carlhubbsi, 11–Peponocephala electra, 12–Phocoena phocoena, 13–Phocoenoides dalli, 14–Pontoporia blainvillei, 15–Stenella attenuata, 16–Tursiops truncatus, 17–Ziphius cavirostris.

Figure S9. As observed for males (figure 4 of manuscript), the marginal posterior distributions of correlations between changes in female pelvic bone size was significantly correlated with shifts towards larger testis size. Under a phylogenetic model of correlated trait evolution (supplementary methods), the size of the pelvic bone is positively correlated with shifts towards larger testes after accounting for body size evolution (red), while rib bone size does not show this correlation (gray). Only sexually mature females were included in this analysis.

Table S1. Individual level data from bone scans. Only sexually mature individuals included here and in all analyses. NA indicates samples that were absent from museum collection or could not be scanned. Numbers presented for bones are centroid sizes. Museum source indicated in specimen column (BMNH=British Museum of Natural History, CCSN=Cape Cod Stranding Network, LACM=Los Angeles County Natural History Museum, MAL=Marine Animal Life, MH=New England Aquarium, MJM=Michael J. Moore, UMA=University of Massachusetts Amherst, USNM=United States Natural History Museum (Smithsonian), UWBM=University of Washington Burke Museum). Specimens in bold were scanned multiple times to assess technical replication (one juvenile not shown).

Table S2. Morphological data gathered from literature for sexually mature males.

Table S3. Posterior means and quantiles of the parameters of the model presented in equations (7) and (8) of supplemental methods, estimated using only bones from adult males.

Table S4. Marginal posterior distributions of correlations, with length fixed, between changes in rib size, pelvic bone size, and testes size, estimated using only bones from adult males.

Table S5. Posterior means and quantiles of the parameters given the dataset consisting only of bones from males for which we have both ribs and pelvic bones.

Table S6. Marginal posterior distributions of correlations, with length fixed, between changes in rib size, pelvic bone size, and testes size, given only data for bones in males for which we have both ribs and pelvic bones.

Table S7. Posterior means and quantiles of the parameters, for bones from females only.

Table S8. Correlations from females: Marginal posterior distributions of correlations, with length fixed, between changes in rib size, pelvic bone size, and testes size, given bones from females only.

Table S9. Distance-based ANOVA of pairwise differences in pelvic bone centroid size. Results from a nested ANOVA performed on the pairwise distance matrix of relative centroid size (centroid size divided by body length). Significance determined with 10,000 permutations, implemented as “adonis (distance matrix ~ species + sex within species + specimen within sex within species)” in the ADONIS function of the Oksanen et al. (Oksanen et al. 2013) package in R.

Table S10. Distance-based ANOVA of pairwise differences in rib bone centroid size. Analysis as described in Table S9.

Table S11. Distance-based ANOVA of pairwise differences in pelvic bone shape. Analysis as described in Table S9.

Table S12. Distance-based ANOVA of pairwise differences in rib bone shape. Analysis as described in Table S9.

NIHMS626667-supplement-Supp_Material.pdf (852.8KB, pdf)

RESOURCES