Why the long, toothy face? A story of plesiosaurian bimodality

On my arrival at the Université de Liège, I was brought into the fold of an ongoing project concerned with the functional morphology and macroevolutionary landscape of short-necked plesiosaurians. Plesiosaurians (or ‘plesiosaurs’ as many of us will have know them growing up) encompass a large number of genera and species of Mesozoic marine reptiles in the superorder Sauropterygia. Some of you avid readers may be surprised to know that the stereotypical image of plesiosaurs as long-necked, short-bodied and small-headed aquatic hunters is not entirely accurate – sure, there were many long-necked, short-bodied, small-headed species in the group, but there were also a large number of plesiosaurs which went along a different evolutionary pathway. It is this different group – the short-necked, massive-bodied, huge-headed group – which I will tell you about today. And just to make it even more interesting, there are two groups of these short-necked plesiosaurs which came to their similar morphologies quite separately and seemingly independently. Cool, huh?!

…oh, and there are lasers…

An excellent example of the two different types of plesiosaurian. On the left a short-necked plesiosaurian (in this case a pliosaurid) attacking a long-necked plesiosaurian (right). Artwork by Bob Nicholls

The study group for our recently published article were short-necked plesiosaurs – from here on I will refer to them as “pliosauromorph” plesiosaurs. The “pliosauromorph” plesiosaurs include two major phylogenetic clades: Thalassophonea, (part of the Pliosauroidea) and Polycotylidae (an offshoot of the long-necked “plesiosauromorph” plesiosaurs). If you are getting lost in the names, bear with me – I will do my best to keep it simple from here on! Over the years, “pliosauromorphs” have in general been clumped together due to their similar appearance, with their morphology suggesting a certain amount of phylogenetic relationship and a generally similar ecology, i.e. “pliosauromorphs” were probably all occupying a similar role (niche) in their ecosystem. Moreover, the clade containing the largest “pliosauromorph” species (Thalassophonea) have been envisaged for many years as apex predators of their time, with genera such as Pliosaurus, Kronosaurus and Sachiasaurus sporting skulls over 2m in length, armed with teeth up to 11cm in length. By contrast, Polycotylidae (i.e. polycotylids) in general exhibit smaller body sizes, narrower skulls and smaller and more numerous teeth. Could this be a classic example of convergent post-cranial body-plan (bauplan) with divergent cranial shape (and associated feeding guild)? For some time this was inferred to be the case. But there was in fact no empirical investigation into this – it has been accepted as being the case without really being tested in a comparative framework. Enter: the SEASCAPE project from EDDy Lab (Université de Liège).

A comparison of “pliosauromorph” plesiosaurs: (from top left) skeleton of thalassophonean Liopleurodon ferox (Tübingen University; image by Wikimedia user Ghedoghedo); skeleton of polycotylid Dolichorhynchops osborni (Kansas University KUVP 1300; image by Wikimedia user FunkMonk); paleoart reconstruction of a pliosaur and polycotylid (copyright Gabriel Ugueto)

Our study as part of the SEASCAPE project set out to quantify cranial and postcranial morphological features of “pliosauromorph” plesiosaurs. Our mission: to develop a protocol to analyse patterns of morphological variation and convergence by using the density of phenotypes projected into a multivariate morphospace to approximate a macroevolutionary landscape. [Sounds complex – let me break it down for you] We aimed to measure multiple features of “pliosauromorph” plesiosaur skeletons (mostly in the skull) which would then be assessed using ordination analyses to explore patterns of variation in functional morphology. In theory, species with similar functional morphologies would group closer to one another in graphical outputs (e.g. principal coordinates analysis). We would then be able to test how many species were occupying specific areas of “morphospace” (i.e. two-dimensional representation of n-dimensional shape space…which in this case boils down to a graph with each point representing a species of plesiosaur). The density of points in a given region of 2D morphospace offers an indication of how frequently this particular suite of morphological characters / functional traits occurs – regions with very high density of points suggests a functional morphology which was evolved over and over [if it ain’t broke, don’t fix it!]. Finally, we hoped to demonstrate how phenotypic densities varied across the phylogenetic tree of “pliosauromorph” plesiosaurs, and see which morphologies evolved repeatedly. Was there only one region of high density, suggesting a single “pliosauromorph” body plan? Were there two peaks (one thalassophonean and one polycotylid)? Or was it more complicated? Here’s how we found out.

Holotype skull of Plesiopleurodon wellesi (CM 2815), laser scanned as part of the SEASCAPE project. Scanned by Rebecca Bennion at the Carnegie Museum of Natural History. Image by Valentin Fischer in ©Meshlab

The method involved the measuring of multiple features on the skull, jaw, and flippers of all known “pliosauromorph” plesiosaurs with remains complete enough for measuring. Some of these specimens were measured from 2D images from published literature or from in-person images…and some were digitised using the EDDyLab Creaform laser surface scanner, which provided some gorgeous models which we will doubtless be using in scientific outreach in the near future! [click here to check them out]. The measurements were then used to calculate functionally informative ratios – for example, the size of the orbit provides a good indication of the visual prowess of the animal, and potentially whether it could negotiate its habitat in low light levels (e.g. Humphries & Ruxton 2002). Unfortunately, not all species had sufficient data available to qualify for further analysis. Our application of a 45% completeness threshold meant that our initial species count of 40 was reduced to 29, which in itself is no small amount (>70%) of the total species coverage for “pliosauromorph” plesiosaurs. We then utilised an up-to-date phylogenetic tree to explore how these functional morphological data were spread across the tree, and whether the morphology matched closely matched intrafamilial relationships; to do this, we used a cluster dendrogram (see below). If phylogeny ≠ phenology, then we would likely be looking at examples of convergent evolution. As it turned out, the cluster analysis hinted at a bimodal (i.e. two-peaked) evolutionary landscape for “pliosauromorph” plesiosaurs, with several taxa which did not fall into the morphological bracket which would have been expected from the phylogeny:

Cluster dendrogram (adapted from Fischer et al. 2020) demonstrating the division between two groups of “pliosauromorph” plesiosaurs (top: longirostrine; bottom: latirostrine) with example taxon bodyplans (artwork by Gabriel Ugueto).
Two highlighted groupings within each cluster flag up the species which are exhibiting divergent cranial morphologies away from their phylogenetic clade (Plesiopleurodon [polycotylid]; Luskhan, Marmornectes, Stenorhynchosaurus, Peloneustes [pliosaurids])

Once we knew there were two main phenotypic groups but with phylogenetic overlap, we conducted a series of statistical tests to tease apart the data and find out exactly what was going on. Firstly, we checked how much statistical support there was for the clusters we recovered in the cluster dendrogram (see above) [spoiler alert, they were significantly different!]. Secondly, we used both principal coordinates analysis (PCoA) and non-multi-dimensional scaling (NMDS) to generate 2D morphospaces accounting for the largest proportion of shape variation in the sample. Using the phylogenetic tree, we converted the morphospaces into phylomorphospaces (see below), and applied a kernel 2D density estimator to project phenotypic densities onto the graph. The result was a 2D phylomorphospace and a 3D density-based macroevolutionary landscape, both demonstrating two distinct adaptive peaks for the “pliosauromorph” plesiosaur dataset [check out cool 2D and 3D manipulations of the landscape here]. These peaks are generated by specific morphologies: a “latirostrine” (broad-snouted) peak and a “longirostrine” (narrow-snouted) peak:

Adaptive landscape of “pliosauromorph” plesiosaurs. A phylomorphospace (left) highlights the presence of almost exclusively thalassophonean taxa (plus Plesiopleurodon) in the “latirostrine” adaptive peak: elongate snouts with broad bases, large teeth and deep lower jaws). On the opposite side, the “longirostrine” adaptive peak (elongate snouts with narrow bases and numerous small teeth) constitutes almost all the polycotylids, but also several thalassophoneans and early pliosaurids. The 3D macroevolutionary landscape (right) demonstrates that there is a greater density of “longirostrine” species (higher peak), whereas the “latirostrine” peak is broader suggesting more morphological variation in this group. Adapted from Fischer et al. 2020.

The formation of macroevolutionary landscapes for “pliosauromorph” plesiosaurs was not the only result to come out of this study – we also investigated morphospace occupation through time, cranial disparity, and several different convergence metrics. [For full details on these methods, I invite you to read the open access article!] Our results are overall very complimentary to one another: overall morphological disparity exhibited within each of the two phylogenetic groups is comparable (no significant differences), although thalassophoneans exhibit greater cranial disparity (potentially due to their greater lineage longevity (84 Ma for thalassophoneans vs 47 Ma for polycotylids). The application of Stayton convergence metrics to a series of inter-clade morphological groupings (e.g. Trinacromerum + Luskhan + Marmonectes; see dendrogram) recovers evidence of significant convergent evolution in the majority of species groupings. Traits of the postcranial skeleton of both groups seem to be evenly distributed, i.e. these do not greatly influence placement in morphospace, nor does it greatly affect phenotypic density. Snout width was a major player in establishing the bimodal landscape, and may also be considered as an ecological signal; the width of the snout provides a proxy for the size of prey that may be ingested, offers a proxy for mediolateral bending resistance (e.g. during tearing or thrashing of prey), and when coupled with snout length offers an indicator of how much water would be displaced during mouth opening / closure. Latirostrine “pliosauromorph” plesiosaurs (by definition) have broader snouts, and would therefore be capable of ingesting larger prey items – the trade-off would be the necessity for greater muscle mass to counteract drag during biting. For longirostrine species, the effect of drag on their long and narrow jaws would have been much lower, and therefore they would have been able to open and close their jaws with less water resistance, albeit to feed on smaller prey items.

A beautiful paleoart depiction of the polycotylid Dolichorhynchops (copyright Brian Engh) hunting for potential prey (fast-moving cephalopods such as belemnites)

All in all, our results recover a pattern of craniodental morphospace occupation of “pliosauromorph” plesiosaurs which is bimodal, being composed of two principal and recurring morphotypes (“longirostrines” and “latirostrines”), which transcend phylogenetic affinities. Transitions between longirostry and latirostry were moderately rare, but involved a significant change of morphology through convergent evolution. This in turn suggests the existence of a pervasive macroevolutionary landscape channeling the craniodental evolution of “pliosauromorph” plesiosaurs – in fact, at this point we should probably accept that our study essentially makes the term “pliosauromorph” plesiosaurs obsolete, because there is no one “pliosauromorph” morphology among plesiosaurs! [#facepalm] Moreover, our results allow us to examine short-necked plesiosaur [it’s back!] bauplan variation through time, and compare our findings with the macroevolutionary histories of other marine reptiles.

Phylomorphospace occupation by different plesiosaur clades through time. Early pliosaurids (red) and thalassophoneans (orange) dominate the Jurassic, with an absence of longirostrine taxa (left side of morphospaces) in the Late Jurassic. Polycotylids (blue) rise to dominance in the Late Cretaceous, coinciding with the decline and extinction of ichthyosaurs. Figure adapted from Fischer et al. 2020

Our method enabled the detection of fluctuations of the (macroevolutionary) landscape through time. The two main morphotypes were already present by the Middle Jurassic (represented by primitive, longirostrine pliosaurids and the earliest latirostrine thalassophoneans). Subsequent evolution repeatedly explored these two morphotypes, indicating the presence of strong and durable morphofunctional constraints. The Late Jurassic marks a chapter in plesiosaur history with no representatives in the longirostrine group (see above); however, this absence may be explained by the dominance of long-snouted ichthyosaurs and thalattosuchians (marine crocodylomorphs) at this time (see Stubbs & Benton 2016). By the Late Cretaceous, the slender-snouted and supposedly fast-swimming polycotylids colonised entirely new regions of morphospace. In actuality, the Late Cretaceous demise of “latirostrine” thalassophoneans resulted in a collapse in the macroevolutionary landscape of short-necked plesiosaurs from a bimodal to a unimodal distribution. The latest Cretaceous short-necked plesiosaurs exploited niches formerly occupied by other highly successful marine reptiles (ichthyosaurs and thalattosuchians), highlighting the effect of early Cretaceous oceanic restructuring as pivotal to the macroevolutionary landscape of plesiosaurs.

But wait! There’s more!

The take-home message of this post should not necessarily be our empirical evidence for the bimodality of short-necked plesiosaur morphology, nor the effect of Cretaceous oceanic restructuring on the evolution of this iconic marine reptile fauna. More importantly, our protocol is easily applicable to any set of taxa for which independent ecomorphological and cladistic data can be gathered – and as such, we hope to see many more applications of this holistic approach to answer questions about evolutionary ecomorphology, phylogenetic influences, and establishment or preservation of macroevolutionary / adaptive landscapes.

This blog has been based on Fischer, V., MacLaren, J.A., Soul, L.C., Bennion, R.F., Druckenmiller, P.F., & Benson, R.B.J. (2020) The macroevolutionary landscape of short-necked plesiosaurians. Scientific Reports 1-12. doi: 10.1038/s41598-020-73413-5

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s