The aim of this project was to explore morphological variation in the elasmobranch braincase using a diverse set of evolutionary analyses, and determine the extent to which braincase shape is a predictor of aspects of elasmobranch life-history and/or ecology in a phylogenetic context. Ecomorphological studies often reason that morphological shape variation under study has some functional significance. At the same time, functional and anatomical studies suggest that specific components of the elasmobranch braincase have different functions [32–34]. Comparative phylogenetic and morphometric analyses across the diversity of elasmobranchs can integrate these insights into a broader evolutionary picture.
We have uncovered several results that may be of evolutionary importance. Neurocranial shape variation is not evenly distributed across the braincase, but is concentrated in the orbits and rostrum, regions of the braincase which previous authors have suggested to be of high functional importance (Fig. 1) [32–34]. Rostrum geometry has obvious hydrodynamic consequences [32, 34], but also serves to support an anterior extension of the pectoral fin web and snout in batoids [33]. Orbit geometry can have profound variation in morphology with a range of functional correlates. These range from the extreme case of the cephalofoil in hammerheads [35] to the role of the orbit wall as a major site of attachment for the upper jaw in squalomorphs [36–37]. Furthermore, orbit diameter and shape is a functional correlate of light conditions [38–40] and therefore likely habitat occupation or life history, which has drawn interest in studies of fossil chondrichthyans [10]. We have found strong evidence for correlation between shape and ecology in the elasmobranch braincase – both at the level of the overall neurocranium, and that of individual ‘modules’ or subsets of neurocranial geometry (Table 1; Table 2). This, together with the results of modularity analyses, points towards the possible presence of independently evolving subunits within the neurocranium, that broadly correspond to the embryonic divisions of the skull (Fig. 5; Table 2).
In the absence of functional studies it is difficult to draw robust conclusions about selection acting on the braincase. But in the following sections we expand upon the most significant correlations (or lack thereof) between shape and ecology/life history in the braincase and comment on potential functional/evolutionary relationships that they imply.
Decoupling of the elasmobranch jaws and braincase reduces constraint and alters ecological signal
The most notable difference between our results and those recovered from other vertebrates is the absence of any significant relationship between neurocranium geometry and diet (Table 1). In other groups prey acquisition/handling is thought to be amongst the most significant selective pressures acting throughout the cranium, and as a result neurocranium shape frequently correlates with aspects of diet and trophic ecology (2,41–42). In this regard the elasmobranch cranium displays mosaic evolution, where different parts of the skull show contrasting patterns of ecological signal. Whilst mandible shape does indeed correlate with some aspects of diet [43], this signal is virtually absent from the neurocranium (Table 1). In most tetrapods there is an intimate musculoskeletal integration between the mandibles and cranium. In teleosts, the maxilla and premaxillar are highly kinetic and the upper jaw and suspensorium can be tightly connected to the braincase. By contrast the elasmobranch cranium is uniquely decoupled, with a few, often highly, kinetic skeletal connections between the braincase and jaws [36–37, 44–46]. Thus we might predict that elasmobranch neurocranial geometry may have a different degree of correlation with diet. However, we caution about generalising too strongly. There are very clear instances, such as in heterodontiform sharks, where major excursions of geometry associate with specialisation in feeding such as durophagy. A lack of clear signal between neurocranial shape and diet could be masked by some of the more extreme shape divergences such as those seen in hammerheads and pristiform sharks.
Compared to many other vertebrate groups the elasmobranch neurocranium is highly integrated and composed of relatively few skeletal units separated by discrete sutures (Table 2; Supplementary Table S7) [3, 47–48]. This suggests levels of correlation between the geometry of different regions of the neurocranium in elasmobranchs is amongst the highest observed in vertebrates. Some clades (such as pelagarian teleosts) display similar levels of integration, however in most vertebrate groups the cranium is comprised of at least five (and in many cases more than 10) discrete modules [3, 47–48]. High levels of integration between subunits complex morphological structures are thought to be underlain by positive trait correlations, enhacing the response to selection and facilitating rapid, correlated morphological evolution [49–51]. The loss of skeletal connections between the two may have released the neurocranium from significant constraint, enabling it to evolve under selective regimes less dominated by prey handling. Evolutionary constraint upon the braincase resulting from articulation with the jaws has been reported in other groups [52]. Indeed, many of the morphological ‘outliers’ amongst elasmobranchs, such as hammerhead sharks, display highly conserved jaw articulation despite exploration of other regions of neurocranial morphospace. Importantly, this does not necessitate those connections between the jaws and neurocranium were lost in elasmobranchs to directly facilitate increased braincase integration – this transition is also cited as having provided elasmobranchs with unparalleled behavioural/morphological flexibility with regards to the feeding apparatus [34]. It is clear however that the anatomical decoupling of the jaws and the braincase in elasmobranchs has enabled both structures to evolve independently, acquiring different patterns of ecological signal in the process.
Relationships between braincase shape and other ecological variables
Herein correlations between neurocranium shape and ecology in Elasmobranchii are restricted to three broad categories (depth, ‘water conditions’ and latitude), with most other correlations being lost when taking phylogeny into account (Table 1; Table 2; Supplementary Table S4). Whilst the absence of rate differences between modules in the selachimorph neurocranium (Supplementary Table S5; Supplementary Table S6) provides rudimentary evidence of natural selection across the braincase [53–54], this does not match observed patterns of ecological signal, which appear to differ significantly between different regions of the braincase (Table 2). As mentioned previously, the orbit and the rostrum were found to be the most variable regions of the neurocranium, which is notable given the hypothesised functional importance of these structures (Table 2; Supplementary Table S7) [32–34]. In the following sections we expand on the three major correlations between braincase shape and ecology, commenting on their potential implications for form-function relationships and morphological evolution in elasmobranchs.
Anterior neurocranium shape in elasmobranchs is correlated with multiple measures of depth (Table 2; Supplementary Table S7). Specifically, depth is correlated with the relative size of the rostral cartilages and orbital processes compared to the rest of the braincase (Supplementary Figures S18-25). Taxa inhabiting deeper waters are characterised by large, robust orbital processes, expansion of the anterior braincase and the presence of substantial rostral cartilages, whereas shallow-water taxa are typified by smaller orbital processes and small or non-existent rostral cartilages (Fig. 4; Supplementary Figures S18-25). Interestingly when comparing between batoids and selachimorphs some differences arise: in the former orbit morphology is not correlated with depth and rostrum morphology correlates only with minimum depth, whereas in selachimorphs maximum depth correlates with orbit morphology and minimum depth is the only depth measure not to correlate with rostrum shape (Table 2). These results are of interest given the high functional importance of both the orbit and the rostrum: the former houses the visual sensory system, whereas the rostrum is the anteriormost portion of the elasmobranch skeleton, and is key to hydrodynamic performance and the distribution of sensory structures such as the ampullae of Lorenzini [32–34]. In some marine vertebrates cranial morphology is known to vary with depth due to its effects on pressure regime and the distribution/sensitivity of sensory structures [55–57]. Large orbits are often seen in deepwater taxa, where they are thought to improve visual performance or represent adaptations to pressure (e.g. scleral capsules). However, a paucity of functional studies comparing kinematics and sensory capabilities between shallow and deepwater elasmobranchs largely precludes further speculation. Nevertheless, there is some correlation between elasmobranch neurocranial morphology and depth, suggesting that depth may be an important driver of neurocranial shape variation, as has been found in other regions of the elasmobranch skeleton [43].
Neurocranium morphology also appears to be influenced by a species’ position on the freshwater-marine continuum. In Selachii (but not Batoidea) all modules were correlated with water parameters (marine, brackish, freshwater) – with correlation being strongest in the occipital module (Table 2). Taxa capable of inhabiting brackish environments generally have dorsoventrally expanded and laterally compressed occipital regions compared to taxa that exclusively inhabit marine environments (Supplementary Figure S7). It is difficult to speculate as to what if any adaptive significance this shape variation may hold, as again the necessary functional studies have not been performed. Theoretically occiput shape could relate to the mechanical properties of the anterior vertebral column, but occiput shape is also likely to correlate strongly with the morphology of the internal semicircular canals. There are likely to be hydrodynamic and sensory differences between marine and brackish environments, but these are difficult to quantify through existing literature, and in any case disentangling how these differences might relate to occiput shape variation observed in this study is impossible at this time. We are also unable to speculate on why this correlation between ability to enter brackish environments and occiput morphology should be restricted to selachimorphs; this would be an intriguing avenue for further study. It is clear however that at least in the case of selachimorph elasmobranchs, ecological characteristics of environments differing in water chemistry may play some role in shaping neurocranial geometry.
Latitude correlates significantly with the shape of the posterior neurocranium in batoids and all modules in Selachii, however this result is difficult if not impossible to interpret on the basis of existing data (Table 2). Species inhabiting tropical latitudes are more likely to exhibit a broad ‘NE’ type neurocranium, with lateral compression and dorsoventral elongation of the occiput, orbit and rostrum associated with higher latitudes (Fig. 4; Supplementary Figure S10-16). Unlike depth and water parameters, correlation between latitude and neurocranium shape is broadly similar across Elasmobranchii, with batoid rostrum morphology being the only module across either Batoidea or Selachii to lack this correlation (Table 2). Biogeography is a significant determinant of the composition of ecological communities [58], potentially enacting various selective pressures upon neurocranial shape. Latitudinal gradients in morphology are known from several vertebrate clades, typically linked to latitudinally-mediated differences in diet or habitat usage [1, 59]. Intriguingly in elasmobranchs, morphologies exhibited at high latitudes appear similar to those found in deepwater taxa and those in brackish environments (Fig. 4; Supplementary Figure S7; Supplementary Figures S10-25). This is indicative of either a monotonic response to temperature, or one-to-many mapping of form to function in the elasmobranch neurocranium. Our cluster analysis provides some evidence for one-to-many relationships, as within each morphotype exist taxa that differ substantially in terms of ecology (Fig. 4; Supplementary Table S1). Even in the case of the OE morphotype – where all taxa use modifications to the anterior cranium to assist in prey capture [34] there is substantial variation in trophic/spatial ecology between taxa (Supplementary Table S1). Unfortunately, a lack of existing literature regarding functional morphology of neurocranium variation in elasmobranchs, the near ubiquity of latitude as a significant covariate of shape, and the rather coarse nature of species-specific ecological information makes inference of the nature of the relationship between neurocranial shape and latitude at best speculative. Nevertheless, it is clear that biogeography – or some correlate of biogeography- is an important factor in shaping elasmobranch neurocranial morphology.
Batoids have more modular braincases than selachimorphs
Whilst the neurocranium of both batoids and selachimorphs displays global integration (Supplementary Table S7), there are marked differences between the two in terms of the number of independently evolving ‘modules’ present (Fig. 5). Selachimorphs appear to possess three neurocranial modules, batoids possess four, and integration between modules is generally lower in the latter (Supplementary Table S7). In batoids the occipital and otic regions (which form a single module in Selachii) are separate, and the nasal capsules contribute to the rostrum module rather than the orbit module (Fig. 5; Supplementary Table S2). In Selachii the jaws are articulated to the braincase through the hyomandibular cartilages and a suspensory ligament joining the anterior neurocranium to the palatoquadrate [60–61]. In batoids this ethmopalatine ligament is lost [44, 62], meaning that the otic region of the neurocranium is more important to jaw articulation in batoids than in selachimorphs. The loss of this ligamentous connection may thus have increased the functional separation of the otic and occipital regions of the batoid neurocranium. In this scenario, increased modularity between the otic and occipital regions is adaptive as it overcomes constraint imposed by subunits of a morphological structure that differ functionally. As a result, batoids have far greater motor control in the feeding apparatus and have evolved many innovations to the hyomandibular musculature to facilitate specialisation [5, 44, 62–63] which may not have been possible in the absence of some degree of evolutionary independence between the occipital and otic regions. The loss of the ethmopalatine ligament would also likely have released the rostrum from evolutionary constraint, enabling greater optimisation of hydrodynamic function and the evolution of novel, batoid-specific roles in prey acquisition [34]. Of course, it is difficult to comment on the ancestral state of modularity in the elasmobranch braincase given these major differences between batoids and selachimorphs. However, it does appear that the loss of the ethmopalatine ligament, combined with increased modularity in the posterior neurocranium has enabled batoids to evolve complex and flexible prey-acquisition strategies that may not otherwise have been possible.
Correlates of evolutionary rate and morphological disparity
We recovered evidence of numerous correlations between aspects of elasmobranch life-history/ecology and both morphological disparity and rates of morphological evolution (Supplementary Table S5; Supplementary Table S6). Intriguingly these correlates of disparity and evolutionary rate differ are far from identical to the correlates of shape itself (Table 1; Table 2; Supplementary Table S5; Supplementary Table S6). Morphological disparity was found to correlate with water parameters, much like some aspects of neurocranium shape (Table 2; Table S5), but also with factors such as reproductive mode and habitat (Supplementary Table S5). This is even more-so true in the case of evolutionary rate, which correlated significantly with almost all tested possible ecological and life-history factors (Supplementary Table S6). Given the large number of correlations present it is extremely difficult to extract any meaningful evolutionary inference from these results alone – although it may well be the case that at least some of these correlates to have a genuine influence over rates of evolution or morphological disparity. For example, rates of morphological evolution in the elasmobranch braincase appear to correlate inversely with latitude such that morphological evolution proceeds more rapidly in the tropics (Supplementary Table S6). This is in-line with the extensively studied latitudinal species diversity gradient, for which much support has been gathered from marine Osteichthyes [64–65]. These results should be taken with extreme caution however. The prevalence of significant correlations between ecology and rates of evolution could equally result from the high dimensionality and low effective sample size of our data. There are no clear and interpretable trends present in our disparity results (Supplementary Table S5),but given that there are various ways of measuring morphological disparity [66] future studies using different measures may uncover different results.
Limitations and future work
There are several limitations to this study, predominantly relating to the range of data available. Future studies should – wherever possible – seek to increase taxonomic coverage, both extant and extinct. Although the preservation potential of the elasmobranch neurocranium is low [67], preserved neurocrania are known and could be included in future studies. Only a single specimen for each species in this study, despite significant intraspecific variation in elasmobranch cranial morphology [68–69] however we considered this approach valid as interspecific morphological differences sufficient to distinguish between species are thought to be present regardless of ontogenetic stage or sex [69]. More significant is that it is difficult to speculate on the selective regimes influencing the neurocranium, or the nature of putative trait correlations and modules, without a priori knowledge of the developmental/genetic basis of morphology [70]. Evo-devo studies (e.g. [71]) focussing on the neurocranium should form the major focus of future studies investigating elasmobranch braincase evolution. Finally, it is important to mention that all phylogenies, models of trait evolution, and calculations of ancestral states are intrinsically hypotheses, subject to revision upon the incorporation of additional data.