Aposematic organisms display warning signals (for example conspicuous colours, distinctive odours or sounds) that inform predators about their possession of secondary defences (such as toxins, spines, irritant hairs, etc.). Predators learn about this association, and eventually learn to avoid defended prey, but this becomes very difficult if the warning signals are variable. The dyeing poison frog (Dendrobates tinctorius) is an example of species with within-population variation in warning signals. During both my doctoral and postdoctoral research I have aimed to understand how this colour pattern variability in aposematic species is maintained.

I have studied the dyeing poison frog at Les Nouragues Reserve in French Guiana since 2009, mostly trying to understand how its phenotypic diversity can persist within a population. I have addressed the possible roles of natural and sexual selection (Rojas & Endler 2013), the link between colour-pattern geometry and behaviour (Rojas et al. 2014a) , the effect of differences in light environment (Rojas et al. 2014) , and spatial patterns of predators and prey (Endler & Rojas 2009) as possible (non-mutually exclusive) explanations for the maintenance of warning signal variation.

However, warning signal variation can also occur between populations of a given species (polytypism). In collaboration with Brice Noonan and JP Lawrence (University of Mississippi, USA), Johanna Mappes (University of Jyväskylä, Finland), Antoine Fouquet (CNRS, France), Ralph Saporito (John Carroll University, USA) and Elodie Courtois (CNRS, French Guiana), I’m trying to get a better understanding of the origin of the phenotypic diversity displayed by the dyeing poison frog across populations, and investigate the possible selective pressures leading to it.

We recently found that new signals can arise in a population which has a strong warning signal that predators find easy to learn and to generalise to other forms, and that is associated with a highly unpalatable chemical defence (which, by the way, is not necessarily one with higher toxin content or toxic compounds richness!). Furthermore, we found that weak warning signals can persist if gene flow from populations with a more powerful warning signals is restricted (Lawrence et al. 2019). See the GIF abstract that my collaborator JP Lawrence did to summarise our results!