Research Undertaken

The Strawberry Poison-dart frog, O.pumilio, lives in aquatic lowland regions on the Caribbean coast of Central America (in particular Nicaragua, Costa Rica and Panama). For most of its range it is found in one colour pattern. This colour is believed to be aposematic – that is, the bright colour serves as a warning to would-be predators that the frog is poisonous. Thus if a predator eats a frog, it will be ill, but will remember next time that bright frogs are not to be eaten. The bright colouration therefore protects O.pumilio. However, in the Bocas del Toro archipelago in Panama there are 15 different colours of O.pumilio seen. Each of these alternative phenotypes occupies a different island (or where two phenotypes occupy the same island there is evidence that the populations have come from different sources and been reintroduced to one another by humans). There are two other species of poisonous Dendrobatid frogs that also inhabit this region, occupying the same habitats (often found within centimetres of one another on the forest floor). Curiously, these species do not exhibit the same phenotypic variation as O.pumilio. Since O.pumilio females do the bulk of caring for the offspring, and this is not the case in the other species, it has been hypothesised that sexual selection may be the force responsible for the wide variation in phenotypes in the focal species as opposed to the others. Using quantitative genetics, Professor Iwasa and I attempted to demonstrate whether or not this was possible.


Figure 1 – O. pumilio, the Strawberry poison-dart frog. This is the general colouration, found throughout most of the frog’s range.

Quantitative genetics is the theoretical study of traits that are controlled by a large number of genes, and are generally continuous in nature. They are also often affected by environmental factors. Examples are height, weight, and, in our case, skin colour. We simplified to consider frog colour as a real-number x, and female preference (the colour of frog they’d most like to mate with) as y. While males express only colour, females have both colour and preference. We defined relative fitness functions for males and females, and from these equations we calculated that in the absence of drift, populations under sexual selection are likely to evolve so that mean colour   and mean preference   are equal. This sort of assortative mating is seen in frog populations in the real world.

However, in our small island populations, random drift will occur. This will happen because in finite populations, chance events will lead to the population means for both colour and preference being slightly different than what they would be from selection alone. Since drift is likely to affect both O.pumilio and the other two species in a similar fashion, we asked the question: what difference does sexual selection make to the variance between different populations caused by drift?

Incorporating existing models for random drift and sexual selection, we established that in some cases the variation in skin colour between sexually selecting populations can be much larger than that between non-sexually selecting populations. However, it depends on the parameters of the model: in some circumstances the variation will be more or less equal in both cases, and in other cases the variation between populations will actually be less in the sexually selecting species!


Figure 2 – graph indicating under what circumstances variance is increased (ie what conditions would be required for sexual selection to explain the variation in O.pumilio). Where the graph is red, variance is increased. Where the graph is blue, it is actually suppressed. The yellow indicates there is little effect on variance.

To cut a long story short, if genetic variance in female preference is greater than genetic variance in colour, and females do not suffer very negative consequences for preferring males of extreme colours compared to what males suffer for bearing those colours, the random drift in female preference can be transmitted to colour through sexual selection. Mean population colour is then pushed around by random drift, and also by drift in female preference. If these parameters apply to O.pumilio, then it is possible that sexual selection could be the cause of the phenotypic variation seen in the Bocas del Toro archipelago. While we cannot be certain about the parameter values in the real world, those required seem plausible at least. This “sex-inflated drift” has not been theoretically described or investigated previously to our knowledge, and may play a role in speciation in a wider range of situations.

Page last modified on 20 jan 10 19:45