Evolutionary transitions between the two major predator avoidance strategies aposematism and crypsis are expected to be associated with changes in many important traits of animals. However, empirical studies on populations experiencing ongoing or recent transitions between these strategies are rare. This thesis investigates the co-evolution of traits among populations of the Strawberry poison frog D.pumilio in Bocas del Toro, Panama. I found that all investigated populations were genetically distinct but that colour and pattern did not correlate with genetic or geographic distance, which suggests that selection needs to be invoked to explain the observed variation. Based on the chromatic contrast between frog dorsal colour and the natural habitat substrates used by the frogs, the populations were defined as bright or dull coloured. I found that frogs from bright coloured populations were larger. This is expected if aposematism is enhanced by large signals while crypsis is enhanced by small size. Further, individuals from bright coloured populations had a coarser black dorsal pattern, which is expected if crypsis is impaired by a bold pattern. The importance of pattern coarseness was confirmed by an avian detection experiment showing that coarse patterned dark green prey were more easily detected than dark green prey without pattern or with fine pattern. I put forward the hypothesis that enhanced protection, gained by aposematism, may affect behaviours that influence dispersal and pairing patterns. Indeed, males from bright coloured populations displayed at more exposed sites and showed a tendency to be more explorative and aggressive. In summary, my results show that the bright and dull coloured populations most likely represent an aposematic and a cryptic strategy, respectively. Furthermore, I show that evolutionary changes between aposematism and crypsis can be associated with coevolution of both morphology and behaviour. I argue that this coevolution may increase the likelihood of both pre- and post-zygotic reproductive isolation. This is because greater phenotypic differences between populations increase the likelihood of selection against badly adapted migrants and hybrids with intermediate traits.

Flamboyant colors are widespread throughout the animal kingdom. While many of these traits arise through sexual selection, bright coloration can also evolve through natural selection. Many aposematic species, for example, use conspicuous warning coloration to communicate their noxiousness to predators. Recent research suggests these signals can also function in the context of mate choice. Studies of warning color evolution can therefore provide new insights into how the interplay of natural and sexual selection impact the trajectory of conspicuous signal evolution. For my dissertation, I investigated the potential for male-male competition to impact the warning color evolution of a species of poison frog. I focused my work on an exceptionally bright and toxic population of the strawberry poison frog (Oophaga pumilio) where males are brighter than females, a classic signature of sexual selection. In Chapter 1, I used theoretical models of predator and frog visual systems to determine which can see the variation in bright warning coloration within this population. I found that birds, the presumed major predator, likely cannot see this variation, indicating that sexual selection can work under the radar of predators in this species. In Chapter 2, I tested the aggressive responses of males using a two-way choice paradigm that manipulated the perceived brightness of stimulus males. I found that males directed more of their behaviors to bright stimulus frogs, and brighter focal frogs more readily approached stimuli and directed more of their attention to the brighter rival. In Chapter 3, I tested the outcomes of dyadic interactions between males of varying brightness and observed male reactions to simulated intruders in their territories. I found that brighter males initiated aggressive interactions with rivals more readily, and brightness asymmetries between males settled interactions in a way that is consistent with classic hypotheses about male sexual signals. In Chapter 4 I sought to describe physiological correlates of male warning color brightness. While male brightness did not co-vary with classic measures of body condition (circulating testosterone and skin carotenoids), it did correlate with toxins sequestered from the diet and thus appears to be a reliable signal of toxicity in this population. / text