Publisher: Federation of European Neuroscience Societies and Blackwell Publishing Ltd
1. Using data from a 9-year study of the great reed warbler Acrocephalus arundinaceus L. carried out in Sweden, I tested the main assumption of the polygyny threshold model (that there is a cost of polygyny) and the main prediction (that this cost is compensated for). The main question was whether a female that chose a mated male (secondary females) experienced the same fitness as a female that at the same time chose an unmated male (primary females). The study includes a total of 192 breeding seasons of 104 individual females of which 40% settled with already mated males. 2. Eight potential correlates of fitness were examined; namely, clutch size, hatching success, hedging success, survival until breeding age, nest survival, mass of fledglings, mass of feeding females and annual survival rate of breeding females. I started each analysis by controlling for four potential confounding variables (year, study site, female age and breeding date). Then I compared the average fitness measure of primary and secondary females. 3. The only detected potential component cost of polygyny was a higher mortality among nestlings if a non-primary position within the harem was retained until hatching. This was supported by a natural experiment. Females that initially chose mated males but which achieved a primary position after failure of the originally primary female's nest enjoyed an increased fledging success, probably because of more male assistance. 4. The number of recruits, controlling for settling dates, for the two categories of females were almost identical (primary females 0 . 53 recruits per year, secondary females 0 . 51 recruits per year). Since secondary and primary females showed similar rates of survival until next breeding season the fitness of the two strategies appears to be equal. These findings strongly suggest that polygyny in the great reed warbler is best explained by the compensation models such as the polygyny threshold model.