Kevin Arbuckle

Many attributes of species may be linked to contemporary extinction risk, though some such traits remain untested despite suggestions that they may be important. Here, I test whether a trait associated with higher background extinction rates, chemical antipredator defence, is also associated with current extinction risk, using amphibians as a model system—a group facing global population declines. I find that chemically defended species are approximately 60% more likely to be threatened than species without chemical defence, although the severity of the contemporary extinction risk may not relate to chemical defence. The results confirm that background and contemporary extinction rates can be predicted from the same traits, at least in certain cases. This suggests that associations between extinction risk and phenotypic traits can be temporally stable over long periods. The results also provide novel insights into the relevance of antipredator defences for species subject to conservation concerns.

Convergent evolution, or the independent evolution of similar traits, has long been investigated and recognised as an important area of research for evolutionary biology. However, as with many areas of comparative biology, new phy-logenetic methods that enhance our ability to study convergence have arisen with greater frequency in recent years. Consequently, we now have a wide range of tools at our disposal and a rapidly developing conceptual framework to guide us in our analyses. This chapter aims to provide a practical guide for those interested in convergent evolution that will enable new entrants to the field to quickly develop a well-rounded research agenda. Although some methods can be performed in other pieces of (stand-alone) software, this guide will focus on the R statistical environment.

Toxic weaponry in the form of venom and poison has evolved in most groups of animals, including all four major lineages of tetrapods. Moreover, the evolution of such traits has been linked to several key aspects of the biology of toxic animals including life-history and diversification. Despite this, attempts to investigate the macroevolutionary patterns underlying such weaponry are lacking. In this study we analyse patterns of venom and poison evolution across reptiles, amphibians, mammals, and birds using a suite of phylogenetic comparative methods. We find that each major lineage has a characteristic pattern of trait evolution, but mammals and reptiles evolve under a surprisingly similar regime, whilst that of amphibians appears to be particularly distinct and highly contrasting compared to other groups. Our results also suggest that the mechanism of toxin acquisition may be an important distinction in such evolutionary patterns; the evolution of biosynthesis is far less dynamic than that of sequestration of toxins from the diet. Finally, contrary to the situation in amphibians, other tetrapod groups show an association between the evolution of toxic weaponry and higher diversification rates. Taken together, our study provides the first broad-scale analysis of macroevolutionary patterns of venom and poison throughout tetrapods.

This thesis aims to improve our understanding of the macroevolutionary implications of antipredator defences, particularly with regard to how defence impacts biodiversity (including both species and trait diversity). To do this I took a phylogenetic comparative approach and used multiple study systems in an attempt to ensure the generality of my work. I begin by investigating how chemical defence and protective coloration influence ecology by testing for life history and ecological correlates of these defences (Chapter 1). Upon finding evidence for an increased niche space in chemically-defended species, and to some degree in conspicuously-patterned species, I explore whether this leads to increased diversification by increasing speciation rates and/or lowering extinction rates (Chapter 2), as also predicted by escape-and-radiate theory (a major and highly influential framework for the macroevolution of natural enemy interactions). Both conspicuous coloration and chemical defence increased speciation rates, but extinction rates were also raised in chemically-defended lineages, leading to a reduction in net diversification. Macroevolutionary extinction rates may or may not be related to contemporary extinction risk, but if they are then there may be conservation implications by allowing prediction of threat status of species with limited direct information. Consequently, in Chapter 3 I asked whether chemically-defended species are more threatened than those lacking such a defence. In accordance with the macroevolutionary results from Chapter 2, I found that chemical defence is indeed associated with a higher extinction risk even amongst contemporary species. In addition to factors that promote diversity, in this thesis I also investigated convergent evolution as a means of constraining diversity of phenotypic traits, using mimicry as a case study for antipredator defences. Many antipredator defences are convergent to some degree, with examples in the repeated evolution of chemical defences and warning coloration as well as independently derived similarity in protective mimicry. However, methods of quantifying the strength of convergent evolution are lacking, not to mention a conceptual framework to define 'strength' in this context, I began by developing a new method to do this which I called the Wheatsheaf index (Chapter 4). Subsequently, I (in collaboration with a colleague, Amanda Minter) also designed software in the form of an R package (called 'windex') to enable user-friendly implementation of the Wheatsheaf index in a familiar statistical environment to many biologists (Chapter 5). In the final data chapter of this thesis, I apply this method in a case study to explore the patterns of phenotypic convergence that result from the evolution of Batesian and Müllerian mimicry complexes. I find that these two types of protective mimicry are generally characterised by convergence in different broad types of traits, but that the specific traits which converge in a given mimicry complex are less predictable (Chapter 6). Overall, this thesis provides novel insights into the evolutionary patterns and consequences of antipredator defences, develops a framework and methods for the analysis of convergent evolution, and suggests further avenues of research for future studies.