Decision-making in variable environments-- a case of group selection and inter-generational conflict?

Suppose organisms need to engage in a particular action exactly once during some fixed period of time. Further suppose they can time this action to optimise their fitness based on the expected current payoff and the probability distribution of later payoffs. For an example we consider the timing of the annual nuptial flight in eusocial insects. Using two population genetics models, we ask whether stochasticity leads to evolutionary conflict between the queen and her offspring. We find that the winning phenotype is independent of who controls the timing. The best response to any non-equilibrium population strategy is the same in both control scenarios, a result that carries over to the diploid case. Although inter-generational conflict is therefore ruled out, the models support a previous observation that at equilibrium some of the offspring have a lower expected payoff than others. By measuring fitness in terms of relative reproductive success, we show that all individuals are in fact equally well off making group-selectionist arguments unnecessary. As such, the models should improve our understanding of the difficult conceptual problem of the unit of natural selection in stochastic environments.