Broadening the angle of view on aposematism: a comment on Skelhorn et al

From bright colors to bad smells, many animals use signals to warn predators of their unpalatability (aposematism; Mappes et al. 2005). The evolution and function of warning signals has attracted attention since the first evolutionary biologists (e.g., Wallace 1867), and it has been a focus of considerable research ever since. Although much work until the last decade or so focused predominantly on the initial evolution of aposematism, and demonstrating that it works, more recent research has paid greater attention to issues such as the mechanisms that make warning signals effective (Stevens and Ruxton 2012). A common feature of many studies of aposematism is the focus on how predators learn to associate prey defenses with warning signals, with much research exploring how the speed and persistence of avoidance learning is influenced by levels of prey defense and signal conspicuousness. In their recent review, Skelhorn et al. (2016) argue we should move away from this general approach to one of more complexity and realism. They make the case that predators do not simply learn to avoid defended prey, but instead make informed decisions as to whether to attack something based on a suite of other information and contextual factors. These may relate not just to prey toxicity but also to the nutritional value of the prey, predator energetic levels, alternative available prey, and much more besides. Skelhorn et al.’s argument is not new, but their current paper explicitly discusses the issues and benefits of such an approach. In the first instance, Skelhorn et al. discuss how, once an association between prey toxicity and warning signal has been made, a suite of factors should affect subsequent predator responses, including the physiological state of the predator, and levels of information predators have regarding aspects of prey nutrition. Likelihood of attack should reflect a (potentially complex) equation of costs of consuming defended prey versus nutritional benefits, weighed up against factors such as the likelihood of future meals and current predator physiological state and nutritional load. Clearly, these factors do not simply reflect what happens after predators have learnt about the prey, but during the learning process too. For example, hunger state should affect the propensity of predators to attack novel prey and overcome dietary conservatism (Marples and Kelly 2001), and potentially the speed of learning. Skelhorn et al. focus primarily on the cognitive aspects of predator behavior and predator physiology, but there are wider factors to consider too. For one thing, studies of aposematism should adopt a more ecologically relevant approach at times. We need a better understanding of how the prey community as a whole affects the way that predators respond to defended species. This will depend on information gathering not just about the prey currently under investigation but also about the wider environment, including availability of other prey and predator communities. So far, comparatively little work has addressed these sorts of issues (but see, for example, Nokelainen et al. 2014). Ultimately, Skelhorn et al. argue that many of the factors that affect how predators respond to defended prey are poorly understood and that these will have a major bearing on warning signal function and evolution. An approach testing these may help clarify inconsistencies in past work. For example, generalization behavior toward defended prey is currently poorly understood, but it affects how predators respond to other aposematic prey, and to mimics. Inconsistencies in past work may partly reflect a frequent failure to account for predator perception, but the factors Skelhorn et al. highlight are likely important too. Another area is in the study of signal honesty and aposematism. Currently, there is much discussion regarding whether the strength of warning signals is an honest indicator of prey defenses. There is some evidence that this may be the case (e.g., Arenas et al. 2015), but also conflicting results and complex theoretical arguments (see Summers et al. 2015), often comprising a range of largely untested assumptions about predator behavior. The idea that predators do not simply learn to avoid prey and subsequently make binary choices is highly relevant here because honest aposematic signaling would predict that predators make informed decisions about whether to attack or not based on a scale of signal conspicuousness and underlying defense. Ultimately, Skelhorn et al.’s message is that we need to move beyond simply considering prey from being “good” or “bad” and more fully appreciate the complexity and spectrum of factors that affect predator responses. This is certainly something the field needs to do more, though testing these issues and interpreting experimental results will present a new suite of challenges.