Sexual dimorphism in skull size and shape of Laticauda colubrina (Serpentes: Elapidae)

Measurements

I measured 60 cleared and dried skulls of Laticauda colubrina (33 females, 27 males) from the collection of the Field Museum of Natural History, Chicago (Appendix 1). However, some specimens have missing or broken bones, and in three cases (two females and one male) the measurements could not be taken from some specific distances (i.e., missing or broken both supratemporals in female #FMNH 236269), and these specimens were removed from some analyses. Data on the snout to vent length (SVL) and body weight (BW) were recorded for 38 specimens (20 females and 18 males), and BW was measured prior to fixation. Female specimens ranged from 605 to 1440 mm SVL, and their skull length ranged from 13.14 to 31.57 mm, and male specimens ranged from 587 to 888 mm and 13.31–24.18 mm respectively; thus this covers the series from subadult to adult specimens. The following skull measurements (Fig. 1) were taken with a digital calliper to the nearest 0.1 mm (raw data are in Data S1): skull length (SL) measured from the tip of the premaxillary bone to the end of basioccipital condyle; skull height (SH) measured at the highest point of the braincase; skull width (SW) at the widest point of the braincase; parietal width 1 (PW1) at the postorbital articulation; parietal width 2 (PW2) at the narrowest point of the parietal bone; parietal length (PAR) measured along the midline; nasal component length (NCL) measured from the tip of the premaxillary bone to the caudal end of the nasal, septomaxillary and vomerine bones and frontal articulation; nasal length (NL) measured between the most rostral and most caudal edges of the nasal bones’ lamina; nasal width (NW) at the widest point of the nasal lamina; frontal length (FL) between the most rostral and most caudal edges of the interfrontal midline; frontal width (FW1) at the fronto-parietal suture; frontal width (FW2) at the narrowest point; palatine length (PLL) from the tip to the most caudal point of the articulation process; Pterygoid length (PTL) from the most rostral point of the palato-pterygoid articulation process to the caudal tip of the bone; pterygoid tooth row length (PTTL); width of the premaxillary bone (PMW); length of the retroarticular process (PRETR) from the quadrate articulation to the caudal tip of the bone; length of the maxilla (MXL); length of the mandible (MDL); length of the in-lever of the mandible (MD2L) from the quadrate articulation to the anterior tip of the bone; length of the dentary bone (DENT) measured along its dorsal (tooth bearing) process; mandibular fossa length (FMDB); length of the ectopterygoid bone (ECT); quadrate bone length (QL); quadrate crest length (CQL); prefrontal bone height (PFH); length of the supratemporal bone (STP).

Of these distances the following are feeding related structures: CQL, DENT, ECT, FMDB, MDL, MD2L, MXL, PLL, PRETR, PTL, PTTL, QL and STP. Others, like nasal width and length, prefrontal height, and frontal width, are related to support of the sensory organs (vomeronasal organ and eyes). NCL is involved both in the supporting the chemosensory structures and feeding (rhinokinetism: Cundall & Shardo, 1995). Premaxilla width, parietal width and skull length, width and height reflect the overall skull form/shape.

Statistical analyses

To test for sexual size dimorphism in body length I ran an ANOVA on SVL. The skull length, width and height (SL, SW and SH respectively) roughly correspond to head dimensions. Thus to find if there is sexual dimorphism in overall skull/head size I ran an MANCOVA with SL, SW and SH as dependent variables and SVL as the covariate.

To analyze potential shape differences separately of size I standardized the raw measurement by transforming into the log-shape ratios (Mosimann & James, 1979). The log-shape ratio is calculated in the following way: the size of the individual is calculated as geometric mean of all measurements. Then each of the measurements is divided by individual “size” (geometric mean) and the result is the shape ratio which is then log-transformed into log-shape ratio. As such the log shape ratios are size-free shape variables (Mosimann & James, 1979; Claude, 2013). I ran principal component analysis (PCA) on the log-shape ratio variables using the covariance matrix. PCs with eigenvalues of 1 and higher have been considered and then tested with MANOVA. Variables with highest loadings (>0.7) on PC 1 that was shown to be responsible for between-sex differences (see Results) were further explored with MANOVA. In order to find if the overall skull shape changes with size and if follows the same pathway in both sexes I ran multivariate regression with subsequent ANCOVA on regression score on log-shape ratio variables (Mosimann & James, 1979). This has been done using the geomorph package (Adams & Otarola-Castillo, 2013; Baken et al., 2021; Adams et al., 2022) for R v. 4.2.2 (R Core Team, 2022).

The allometric coefficients were estimated from the regression of log-transformed skull measurements against the log-transformed skull length (SL) and log-transformed snout to vent length (SVL). These two different baselines were considered because the SVL itself may be sexually dimorphic, and relying only on this size measure may produce skewed results and interpretation. Because the measurement error was present in both the x- and y-variables the slopes were calculated using Reduced Major Axis Regression (Sokal & Rohlf, 1995). Although the Type II Regression Model (or Major Axis Regression) may seem to be more appropriate in estimating the allometric slopes (see Warton et al., 2006), I decided to use RMA Regression as it allows direct comparisons with the similar studies on other species (Hampton & Moon, 2013; Hampton, 2014; Hampton & Kalmus, 2014; Borczyk et al., 2021; Patterson et al., 2022). All calculations were done using Statistica and software by Bohonak & van der Linde (2004).

Allometry

The allometric coefficients for the studied traits are summarized in Figs. 4 and 5 (confidence intervals and determination coefficients for the allometric equations are presented in the Tables S2 and S3). Skull measurements (SL, SH,) scaled with slightly negative allometry with respect to SVL, with the exception of SW, which showed isometry (females) or slightly positive allometry (males). When scaled against SL most of the characteristics showed positive allometry or isometry, and only PW2 showed negative allometry. Table 3 summarizes the allometric coefficient of other snake species from published studies that used homologous skull distances.

Skull size and shape

Although Camilleri & Shine (1990), Shetty & Shine (2002) and Shine et al. (2002) previously reported that female sea kraits have relatively longer and wider heads this was not reflected by the overall skull shape and size in the current study; in this case the roughly corresponding traits are MDL (for typical head length) and SW (for head width). Both male and female skulls were uniform and their overall proportions changed in the same way when scaled against both SVL and SL. There may be two explanations for these differences. First, my measurements were not homologous with those of the above mentioned authors. For example, the “head width” as measured in previous studies reflects the widest overall point of the head, and thus includes the skull width increased by the quadrate projection and mandible, which extends posteriorly, plus soft tissues (jaw adductor musculature, salivary and venom glands and skin with scales). Jaw adductor musculature increases with positive allometry with respect to skull length (Vincent & Herrel, 2007), and thus it may contribute to the overall head width dimorphism. In contrast, my skull measurements were taken at well-defined points (e.g., at bone sutures), and were not necessarily in the same plane as the widest head point and of course, did not cover the soft tissues.

Although females and males did not differ in overall skull shape (as reflected by SL, SH and SW), specific bones within the skull did show significant sexual dimorphism. The sexually dimorphic differences that are female-biased are those related to feeding efficiency, which may be attributed to sexual and ontogenetic differences in feeding behaviour (from muraenid eels in males and young snakes to larger congerid eels in females). However, Camilleri & Shine (1990) found no sexual differences in the so-called “trophic structures” of sea kraits, and asserted that “sea snakes did not show significant differences in the lower jaw although observed differences were in the predicted direction”. A possible explanation for this discrepancy is the relatively small sample size in Camilleri & Shine (1990), comprising 13 males and 13 females, which may have led to type II error.

Male-biased dimorphism in head shape is restricted to the frontal (FL) and parietal (PW1 and PW2) regions. These are the skull parts that contain two of the major sensory structures (olfactory bulbs and eyes). Pheromone-driven mate searching is well documented in snakes (Mason & Parker, 2010) and may explain male-biased dimorphism in the region bearing the chemosensory structures. However, the use of visual cues in mating behaviour by snakes is largely unstudied. At least in some species males use visual stimuli when searching for a mate (e.g., Thamnophis sirtalis: Shine et al., 2005; Emydocephalus annulatus: Shine, 2005). In addition, sexual dimorphism in eye size has already been reported in numerous snake species (Faiman et al., 2018). Thus male-biased sexual dimorphism in the orbital region may be controlled by sexual selection and related to more effective mate searching by males whereas female-biased sexual dimorphism in feeding-related structures may be driven by natural selection (ecologically-based sexual dimorphism).

Skull allometry

Skull length, height and width exhibited negative or isometric allometry when compared to SVL, as observed in other snake species (e.g., Hampton, 2014; Hampton & Kalmus, 2014; Murta-Fonseca & Fernandes, 2016; Jayne et al., 2022); however, the slopes are remarkably higher in the sea krait compared to the other snakes (Table 3). For example, in the natricine Nerodia fasciata (Hampton, 2014) and Farancia abacura (Hampton & Kalmus, 2014), slopes for SL were 0.62 or 0.56 respectively vs. 0.9–0.93 in this study (Table 3). Thus, although all species show negative allometry of the head with respect to SVL, in sea kraits the relative skull length is only slowly shrinking with respect to SVL. Moreover, the suspensory elements (QL and STP) and jaw and palate bones (MXL, MDL, PTL, PLL) in Laticauda grow with positive allometry with respect to SVL, whereas in other species for which comparable data are available they show negative allometry, with exception for QL in Boiga irregularis (Table 3; Hampton & Moon, 2013; Hampton, 2014; Hampton & Kalmus, 2014; Borczyk et al., 2021; Jayne et al., 2022; Patterson et al., 2022). This increase in the growth rate of structures directly influencing gape size and prey swallowing efficiency may underlie the adaptation to piscivory and aquatic habits in Laticauda. Specifically, an aquatic lifestyle promotes selection for rapid swallowing of prey and is often reflected by the elongation of suspensory elements that contribute to gape size and increase anteroposterior mandibular excursion during feeding (Savitzky, 1983). It is also possible that such a growth pattern is constrained by the space for venom glands. However, to safely evaluate the impact of feeding habits on the allometric patterns of skull bones, taxonomically broader sampling is needed, since some of the differences may reflect phylogenetic, ecological or morphological constraints.

Structures associated with feeding performance, namely the jaws, palatine and pterygoid bones and suspensory elements, scale with positive allometry with the skull length. This is because bigger snakes can usually eat relatively larger prey items and better handle prey items (e.g., Vincent et al., 2006). Relative elongation of the structures that influence the gape size (jaws, quadrates and supratemporals) and help with prey handling (maxilla and dentaries) may be a morphological response to increasing prey size. A similar pattern was found in the natricine Nerodia sipedon, where a dietary shift from fishes to anuran prey has been observed (Vincent & Herrel, 2007), in Nerodia fasciata (Vincent et al., 2007; Hampton, 2014) and in the elapid Pseudonaja affinis (Patterson et al., 2022). In the elapid Notechis scutatus it has been shown that a rapid shift in trophic bone size occurs in response to larger prey (Ammresh et al., 2023). In sea kraits this may be explained as an adaptation to a dietary shift from smaller muraenid eels to bigger congerid eels (Shine et al., 2002; Shetty & Shine, 2002). However, because a similar pattern is also observed in Aipysurus eydouxii, a fish-egg eater with no prey-size constraint in feeding morphology (Borczyk et al., 2021), it may also simply be a functional and structural consequence of size increase (see Bonduriansky, 2007).

Interestingly, with the exception of PW2, all traits that showed sexual differences in growth trajectory are feeding structures. Quadrate length is one of the factors responsible for the maximum gape size, thus limiting the maximum prey diameter that a snake is able to swallow (e.g., Gripshover & Jane, 2021). The palato-pterygoid bar plays a role in prey transport, as the movements of this complex are responsible for pushing the prey into the oesophagus during the oral and oro-cervical phases of prey transport (the so called ‘pterygoid walk’; Kley & Brainerd, 1999). Because adult females feed mostly on congerid eels whereas males prey on smaller muraenid ells (Shine et al., 2002; Shetty & Shine, 2002), these differences in the growth trajectory of the quadrate length (faster in females) may be an adaptation to feeding on bigger prey.

When looking at the skull in general, there was also an interesting pattern of changes in proportions between the anterior-middle-posterior skull segments. The nasal component (NCL) grows with relatively high positive allometry (the slope is equal to ∼1.3) and the same is true for frontal length (FL, slope approximately 1.2–1.3), but the parietal component of skull length (PAR) grows almost isometrically (1.0–1.1, see S2). These distances all contribute to the skull length and show that during snake ontogeny the anterior (preorbital and orbital) region grows faster than the posterior one, i.e., bigger kraits are relatively longer-snouted than their smaller companions. This largely fits the general pattern observed among tetrapods (Emerson & Bramble, 1993) and snakes seem not to be an exception (e.g., Murta-Fonseca & Fernandes, 2016; Borczyk et al., 2021; Patterson et al., 2022). However, at finer scale, this growth trajectory may again underlie the adaptation to aquatic feeding and locomotion, as it has been shown that aquatic snakes have longer and narrower snouts compared to their terrestrial relatives, which reduces the hydrodynamic drag when striking (Vincent et al., 2009; Segall et al., 2016; Silva et al., 2018). A similar pattern of regional changes in skull proportions was found in semiaquatic Hydrodynastes gigas (Murta-Fonseca & Fernandes, 2016): the nasal component has been strongly elongated (from 15.5% of SL in small snakes to 33.9% of SL in large ones), whereas the frontal and parietal components have shrunk (ca. 3–5% of SL). This pattern therefore suggests a consistent developmental trend among aquatic and semi-aquatic snakes; however, at this point there is not enough data on terrestrial and aquatic close relatives to fully compare the ontogenetic allometric (slope changes) patterns of skull growths in ecological and phylogenetic framework.

Because of the magnitude of differences (in slope coefficients or skull shape), studies that used combined samples (e.g., Rossman, 1980; Hampton, 2014; Hampton & Kalmus, 2014; Murta-Fonseca & Fernandes, 2016), although probably correct at the general level, must be interpreted with caution. Females are not just scaled males (and vice-versa) and even subtle shape differences may be underlain by different allometric trajectories (e.g., Wilson et al., 2022) However, the availability of osteological material for such studies is generally limited and does not necessarily correspond with researchers’ needs in the context of taxonomic and ecological diversity (Bell & Mead, 2014).