Thesis title: Macroevolutionary trends and ecological correlates in the craniomandibular complex of carnivorans
Chapter 1
Evolutionary trends (ETs) are traditionally defined as persistent and directional changes in the state of one or more quantitative traits, resulting in substantial changes through time and representing the primary phenomenon characterising evolution at high taxonomic levels and over long-time scales. The first groundbreaking investigations concerning phenotypic directional evolution led to the description of iconic and pervasive trends throughout the history of a single clade, frequently relying on the evidence available from fossil record (e.g., increasing hypsodonty and gradual acquisition of the monodactyl posture in equids caused by the spread of grasslands). Then, the formulation of the so-called ‘biological rules’ extended the definition of ETs to include directional responses to ecological, climatic or biological gradients, with several clades being simultaneously analysed obtaining the same evolutionary pattern (e.g., latitudinal or elevation gradients, like in Allen’s or Bergmann’s rules). ETs constitute an ideal case study to describe and separate tempo and mode in evolution, and their mutually independent variations. The strength of an ET can be quantified as the magnitude of a vector, providing a practical way of representing the speed (or tempo) of evolution. The direction of the same vector represents the mode of evolution (i.e., observed pattern of variation). Phenotypic ETs were frequently found as the resulting evolutionary outcome in studies about the controversial field of evolutionary predictability, that focuses on the occurrence of repetitive and foreseeable patterns in evolution under specific conditions (e.g., recurrent patterns in insular systems). Although the search for new ETs and the validation of the existing ones remain central topics in evolutionary biology, the formulation of a well-supported and operative definition of ET remains an unsolved issue. After introducing a new operative definition of ETs, the main goal of this Chapter was to propose a modern theoretical framework for the study of phenotypic ETs, based on Seilacher’s theory of morphodynamics, that explored the interactions between ETs and the main evolutionary factors involved in their occurrence (i.e., phylogenetic history, developmental constraints, environment, and biological function).
The proposed operative definition of ET was a macroevolutionary transposition of the concept of ‘(non) parallel evolution’ (i.e., continuum from convergent through parallel to divergent evolution) that was originally suggested in order to describe the independent evolution of replicate populations, in reference to bacterial cultures of laboratory experiments. Four possible classes of ETs (i.e., simple trend, parallel evolution , convergence, and divergent evolution) are conceivable from this perspective. The criterion of distinction between them relies on the orientation of the evolutionary trajectories in the multivariate trait space shown by the single clades. Therefore, parallelism is the resulting outcome when evolutionary trajectories point in the same direction, otherwise, a condition of convergence or divergent evolution occurs, depending on whether trajectories respectively point ‘at’ or ‘away from’ a region of the trait space. According to our definition of ETs, parallel evolution, convergence, and divergent evolution can be considered as patterns produced whenever, analysing multiple clades at the same time, a portion of the
groups undergoes an episode of directional evolution, whereas other groups are subject to a different evolutionary regime.
Research on ETs would take a major step forward if a pattern-based approach, like the one described above (that can only validate or disprove the presence of a pattern), was systematically accompanied by multifactorial analyses. These analyses would also allow researchers to take into account data relative to underpinning dynamics leading to the occurrence of directional evolution (e.g., ecological variables or life-history traits). When it comes to evolutionary dynamics occurring in phenotypic evolution, the tangled nature of interactions between the factors underpinning morphological evolution has already been explored by the theory of morphodynamics, concluding that four components have a predominant role: phylogenetic history, evo-devo constraints, environment, and biological function. As anticipated above, in this Chapter we explored the interactions between phenotypic ETs and these evolutionary factors (starting with evo-devo constraints and environment, then shifting onto phylogeny and finally focusing on biological function in the light of new perspectives in morphological quantification), also mentioning applications or methodological approaches that can be used in these contexts. For instance, when it comes to ecomorphological studies on ETs, we suggested that the inclusion of spatially structured variation in evolutionary models (i.e., Wrightian view of evolution) would allow considering metapopulations composed by interacting demes characterised by different trait values: this would enable the distinction between patterns produced by spatial processes, like drift and habitat-specific selection, and ETs associated with non-neutral and directional evolutionary regimes. We also discussed how assemblage-based studies on ETs struggle to describe evolutionary dynamism over large spatial and/or temporal scales (e.g., inconsistent presence of Bergmann’s rule in different assemblages of the mammalian order Carnivora), suggesting to always adopt a phylogenetically-informed approach for the study of macroevolutionary patterns and processes, specifically relying on phylogenetic comparative methods (i.e., statistical models that estimate the evolutionary regime that best approximates the tempo and mode of evolution acting on the considered traits, allowing researchers to correct for biases due to the non-independence of sampled observations in macroevolutionary samples). Finally, we described an intriguing new frontier of macroevolution that is represented by the possibility to perform research on ETs in the presence of extreme environmental shifts, like those resulting from the current phase of climate change, allowing biologists to refine predictions of future or hypothetical evolutionary outcomes and potentially leading to indirect implications in climate change and conservation biology.
Chapter 2
Currently available methods to explore evolutionary convergence either rely on the analysis of the phenotypic resemblance between sister clades as compared to their ancestor, fit different evolutionary regimes to different parts of the tree to see whether the same regime explains phenotypic evolution in phylogenetically distant clades or assess deviations from the congruence between phylogenetic and phenotypic distances. The new R function search.conv allows researchers to test for the occurrence of
morphological convergence in multivariate shape data working with either ultrametric or non-ultrametric phylogenies. It relies on θ, that is the average angle between the phenotypic vectors of putatively convergent species in the multivariate shape space (i.e., a measure of the resemblance between the phenotypes), whose cosine represents the correlation coefficient between these vectors. The simulations performed on this R function correctly identified convergent morphological evolution in 95% of the cases and type I error rate was inferior to 6%. Among the considered real case-studies, convergence tests performed on mandibular shape data of fossil and living felids and barbourofelids using the R function search.conv confirmed that “true” sabertooths (i.e., Homotheriini and Smilodontini) independently converged on barbourofelids in their mandible morphology, whereas Metailurini (i.e., “false” sabertooths) were not found to converge on barbourofelids, as expected.
Chapter 3
The graphical representation of evolutionary trends might be a crucial tool for improving their description and comparing the outcomes of different studies in this evolutionary field. The new R function conv.map allows researchers to visualise the regions of 3D model that underwent convergent evolution and assesses convergence by testing whether phenotypes in distant clades in a phylogenetic tree are more similar to each other than expected from their phylogenetic distance (i.e., adopting a method similar the one used in the R function search.conv). Cranial 3D models belonging to different metatherian and placental clades including sabertoothed species (e.g., felids, barbourofelids, dasyuromorphians) were compared using the R function conv.map in order to validate the presence of convergent evolution and graphically visualise the convergent morphological regions of their crania. Our results revealed a range of shared anatomical features in the premaxillary area, the carnassial region, and in the occipital region around the nuchal crest, which were common to all sabertooth carnivores despite considerable phylogenetic distances. This strongly suggests that, despite some anatomical differences and possible functional diversification within sabertooths, their morphotype universally confers a broadly comparable capacity to hunt and rapidly kill relatively large prey by applying a stabbing bite to the throat assisted by powerful neck muscles, although this specialization may have led to their extinction at different times and locations when large prey became less abundant.
Chapter 4
The reconstruction of ancestral states is a fundamental step for macroevolutionary analyses and it is a required step for applying phylogenetic comparative methods. The inclusion of fossil phenotypes as ancestral character values at nodes in phylogenetic trees is known to increase both the power and reliability of phylogenetic comparative methods applications. A new implementation of the R function RRphylo (named RRphylonoder), that is based on phylogenetic ridge regression, enabled the possibility to integrate fossil phenotypic information as ancestral character values in morphological analyses. Results from simulated data proved RRphylo-noder to be slightly more accurate and sensibly faster than Bayesian approaches, and the least sensitive to the kind of phenotypic pattern simulated. Furthermore, the use of fossil phenotypes as
ancestral character values noticeably increased the probability to find a phenotypic trend through time when it applies to either the entire tree or just to specific clades within it. Then, RRphylo-noder was used on real case-studies, including the evolution of body size (i.e., validation of the evolutionary trend known as Cope’s rule) in caniform carnivores. By regressing RRphylo-noder ancestral estimates against their age our results found moderate evidence in favour of Cope’s rule in caniforms. Although this result is not a robust indication of Cope’s rule per se, it is interesting that the pattern appears even at the level of ancestors with RRphylo-noder.
Chapter 5
Analyses performed in order to test for the presence of convergent evolution in the crania and mandibles of living carnivorans pointed out that this phenomenon appears to be rare within this clade. In particular, almost none of the tested dietary categories reached statistical significance, indicating that the mere influence of diet is unlikely to produce convergent patterns in the cranio-mandibular shape evolution of carnivorans. By contrast, a limited number of cases concerning either ecologically equivalent species or ecologically similar species of different body sizes (i.e., red fox vs Malayan civet, giant panda vs red panda, and
raccoon dog vs raccoon) were found to have convergent cranio-mandibular shapes, potentially suggesting the occurrence of a complex interplay of one-to-many, many-to-one, and many-to-many relationships taking place between ecology, biomechanics, and morphology. The analytical framework relying on multiple convergence metrics with different biological meaning (i.e., C1, θ, and Wheatsheaf index) also allowed to discriminate between episodes of convergence and cases of evolutionary conservatism (i.e., closely related species more similar than would be expected based on their phylogenetic relationships). In particular, our results supported the presence of conservatism within omnivore carnivorans. This pattern derives, in all probability, from the existence of an omnivore adaptive zone in the craniomandibular shape evolution of living carnivorans, with other specialized species emerging from this region of the multivariate shape space.
Chapter 6
Ecomorphological analyses were performed on 2D cranio-mandibular size and shape data of living carnivorans in order to clarify the impact of multiple ecological factors (e.g., diet, aquatic vs terrestrial habitat) on morphological evolution within this clade. To do so, complex simulations were implemented in order to estimate the evolutionary scenario that best reflects the evolutionary history of the carnivoran cranio-mandibular complex, and to assess potential biases resulting from the estimation of the best model of evolution. Our simulations confirmed the need to exercise caution in the estimation of the best evolutionary model for highly dimensional shape data (i.e., few species and many variables) since in these cases the adopted model selection criteria might erroneously prefer the most complex models. Our results clearly highlighted that the invasion of aquatic niches produced a trend towards larger cranial dimensions, with pinnipeds being consistently larger than terrestrial carnivorans due to thermoregulation, basal metabolic costs, and food intake functions imposed by the aquatic environment. When dietary categories were
concerned, highly demanding feeding imposed by the consumption of molluscs resulted in exceptionally high disparity and evolutionary rate for both cranial and mandibular shape, whereas the other aquatic diets equally showed higher levels of disparities, but relatively slower evolutionary rates.
Chapter 7
CRaniofacial Evolutionary Allometry (CREA) is a recently formulated evolutionary trend proposing that, among closely related species, the smaller sized of the group would appear paedomorphic with proportionally smaller rostra and larger braincases. Analyses performed on living and fossil felids and relying on 3D geometric morphometrics, together with phylogenetic comparative methods, pointed out that sabertoothed cats (i.e., Machairodontinae) and big conical-toothed cats (i.e., Pantherinae) represent exceptions to this biological rule since they show a decoupling of the allometry-driven pattern of morphological integration between the rostrum and braincase typical of CREA. By contrast, allometric tests performed within either small conical-toothed cats (i.e., Felinae) or the whole family Felidae reached statistical significance, clearly indicating the occurrence of CREA within these groups. Overall, the adoption of different landmark configurations, phylogenetic hypotheses, and corrections for phylogenetic effect had a limited impact on CREA pattern recognition within felids. These findings disproves the hypothesis that the cranial shape evolution of sabertoothed cats results from a mere case of cooptation and extension of the allometric trend observed in conical-toothed cats. More importantly, these results also suggest that the acquisition of extreme features concerning evo-devo constraints, biomechanics, and/or ecology (e.g., adapting biomechanically demanding structures such as sabertoothed upper canines, occupying extremely narrow and specialised ecological niches) is likely to represent a preferential way to escape from common evolutionary patterns of morphological variation such as CREA, but this is likely to be frequently achieved at the cost of higher extinction rates as also suggested by the absence of CREA in many extinct lineages of non-mammalian synapsids.
Conclusions
The present Thesis confirms that carnivorans represent an ideal case study for analyses concerning morphological directional evolution and phenotypic evolutionary trends in general. The inclusion of fossil morphologies constitutes one of the most valuable addition to many of the analytical frameworks adopted in the previous Chapters and strongly impacts the results concerning the presence and strength of evolutionary trends in morphological data. Similarly, the high taxonomic coverage that characterise many of the morphological samples used in the present Thesis is likely to have greatly improved the ability to validate/confute the occurrence of evolutionary trends within the considered samples. Future advances in research on phenotypic evolutionary trends, and macroevolutionary pattern in general, are expected from the implementation of a complex ecological modelling in the study of morphological evolution. Linking these biological fields is also suggested to potentially provide useful information for investigations on conservation prioritization, phenotypic diversity, and predictability of evolution in the near future. Similarly, combining
modern techniques of morphological quantification (e.g., geometric morphometrics and finite element analysis) with advanced methods in ecological niche evaluation (e.g., species distribution modelling) is supposed to be a promising way to bring deep insights in the understanding of ecomorphological pattern and processes. Hopefully, the operative definition of evolutionary trend provided in Chapher 1 and the analytical approaches adopted throughout the present Thesis will represent a good starting point for future pattern-based studies on carnivoran evolution, and macroevolution in general, enabling in such a fascinating topic of evolutionary biology, figuratively speaking, “new avenues for old travellers”.