• colour polymorphism;
  • habitat choice;
  • natural selection;
  • predation;
  • sexual dimorphism

Populations of pygmy grasshoppers, Tetrix subulata, exhibit genetically coded discontinuous variation in colour pattern. To determine whether the dynamics of this polymorphism is likely to be affected by selective processes, rather than by stochastic events, we experimentally manipulated colour patterns of free-ranging grasshoppers and then calculated and controlled for differences in capture probabilities between categories of individuals before estimating and testing for differences in survival using mark–recapture data and program SURGE. We found that paint treatment had a significant effect on survival, and that the relationship between colour pattern and survival was different in males and females. Our analyses also revealed significant differences between sexes in relative frequencies of natural colour morphs, body size, activity pattern, dispersal distance and microhabitat use. These findings accord with the hypothesis that colour pattern and behaviour jointly determine susceptibility to visual predators. Our data enable us to reject the null-hypothesis that colour pattern is a selectively neutral character and that the polymorphism is maintained solely by stochastic processes, such as random genetic drift and founder events. Indeed, the effect of dorsal coloration on survival, together with associations between colour pattern and many biologically important traits (body size, behaviour, thermal capacity, physiology and reproductive performance), suggests that colour pattern is likely to significantly influence individual fitness, and that the polymorphism must be maintained by some active process, such as spatially variable selection in combination with gene flow. The possible role of colour polymorphism as an intermediate stage in the evolution of sexual dichromatism in animals is discussed.