Batesian mimicry is an interspecific relationship in which a palatable species, the mimic, avoids attacks from predators by resembling an unpalatable species, the model. Mimicry has long been studied to understand the evolutionary dynamics of adaptation, yet many factors affecting the co-evolution of mimics and models are understudied in natural systems. In the first portion of this dissertation, I describe mimicry between two salamander species in which the erythristic color morph of Plethodon cinereus (the mimic) resembles the juvenile eft stage of Notophthalmus viridescens (the model). I found that the coloration of mimics resembles that of models, particularly from the perspective of avian predators. I also discovered that while mimetic phenotype appears to have converged on model phenotype, selection from predators may drive models to appear distinct from mimics, particularly by appearing more conspicuous against common backgrounds.

In the second portion of this dissertation, I evaluate how variation in model toxicity may influence the evolution of mimicry by measuring the concentration of the neurotoxin, tetrodotoxin, in N. viridescens individuals from localities with varying degrees of mimicry. I found that despite significant variation in toxicity among localities, model toxicity did not predict either the occurrence of mimicry or variation in mimetic phenotype. These findings suggest that even though theoretical and experimental studies have predicted a link between model toxicity and mimicry, this relationship may not always be strong enough to influence mimic evolution in natural systems.

In the final portion of this dissertation, I identify selection on P. cinereus coloration by mammalian predators. I evaluated mammalian selection on P. cinereus color morphs by measuring mammal attacks on clay replicas of P. cinereus and comparing these attacks against predictions of predator behavior under different hypotheses. Intriguingly, selection by mammalian predators was inconsistent with predictions from the hypothesis of mimicry. Instead, after developing a likelihood-based method for combining non-exclusive hypotheses of predator behavior, I found that mammals appear to use visual cues from salamander prey to preferentially attack familiar and conspicuous individuals. These results are important for the evolution of coloration in P. cinereus because they show that multiple predator species may differentially influence the evolution of coloration and thus complicate predictions of mimicry evolution.