The assumption that the colours of the green poison frog, Dendrobates auratus, serve as a warning of unpalatability is not supported because the requirements of the theory of aposematism are not met. The theory predicts that the aposematic coloration be conspicuous to potential predators, yet this cannot be demonstrated. The only predator found for D. auratus was the theraphosid spider, Sericopelma rubronitens, which hunts using vibrational cues. Only after sampling all frogs offered did spiders reject the toxic D. auratus over nontoxic Physalaemus pustulosus. The spider's decision to reject D. auratus as prey was made without visual input, and therefore the colours and toxicity of the frogs are decoupled with respect to this predator. Discovering that the crab Armases angustum is a predator of D. auratus tadpoles suggests that D. auratus is most vulnerable to predation during the nontoxic tadpole stage. Indirect methods of estimating predation were unsuccessful. The use of plasticine model frogs to estimate avian predation yielded no useful information and although traumatic injury is usually attributed to attempted predation, this is not true for tropical anurans. The levels of traumatic injury for D. auratus and a cryptic nontoxic frog Physalaemus pustulosus were similar when the life span of the frogs was taken into account. The traumatic injuries in these frogs are a function of interactions with small invertebrates that are neither predator nor prey. The theory of aposematism also predicts that there should be low pattern variation within an aposematic species so that a single search image can be formed and avoided. The patterns of D. auratus are individually distinct and there is significant interpopulational variation in appearance due to differences in colour, pattern and animal size. Selection for conspicuous visual intraspecific communication may have resulted in these differences in appearance. Mitochondrial 16S rRNA and cytochrome b sequence data ind

No description available.