How should we define fitness in structured metapopulation models? Including an application to the calculation of evolutionarily stable dispersal strategies

We define a fitness concept applicable to structured metapopulations consis ting of infinitely many equally coupled patches. In addition, we introduce a more easily calculated quantity R-m that relates to fitness in the same m anner as R-o relates to fitness in ordinary population dynamics: the R-m of a mutant is only defined when the resident population dynamics concierges to a point equilibrium and R-m is larger (smaller) than 1 if and only if mu tant fitness is positive (negative). R-m corresponds to the average number of newborn dispersers resulting from the ton average less than one) local c olony founded by a newborn disperser. Efficient algorithms for calculating its numerical value are provided. As an example of the usefulness of these concepts we calculate the evolutionarily stable conditional dispersal strat egy for individuals that can account for the local population density in th eir dispersal decisions. Below a threshold density (x) over tilde, at which staying and leaving are equality profitable, everybody should stay and abo ve (x) over tilde everybody should leave, where profitability is measured a s the mean number of dispersers produced through lines of descent consistin g of only non-dispersers.