As was the case for first-year birds during winter, adult males from larger colonies were favored in cooler and wetter years during the breeding season, whereas the reverse held for birds from smaller colonies ( Fig.
For breeding adult males, annual variation in the directional selection coefficient was significantly predicted by extent of drought in Nebraska, but not by drought in Argentina. Birds from larger natal colonies were favored in cooler and wetter years on the wintering range, whereas those from smaller colonies did better in hotter and drier years ( Fig. Results and Discussionįor first-year birds, variation in the annual directional selection coefficients was significantly predicted by extent of drought in Argentina, but not by drought in Nebraska. The wintering range in northeastern Argentina also exhibits high annual variability in the degree of drought ( 35). Earlier work showed that these birds are sensitive to drought conditions that influence breeding time and that the effect of colony size on breeding time varies with the extent of summer drought ( 34). To determine whether any annual changes in selection were related to seasonal weather conditions, perhaps through effects on parasite populations or food supply, we examined relationships between selection gradients and drought indices for the birds’ 3-mo (May–July) breeding period in Nebraska and for the 3 mo (November–January) they are resident in Argentina. Analyses for adults used multistate models ( 33) with two states, in which (for computational reasons) survival was estimated for each focal colony size as one state and all other colony sizes as a second state. The regression for first-year survival included the effect of hatching date, a major determinant of first-year survival in many birds ( 32). Linear selection gradients describe positive or negative directional selection, whereas quadratic ones were indicative of stabilizing (if negative) or disruptive (if positive) selection on colony size. We used the mark–recapture data ( Table S1) with Program MARK ( 29) to perform linear and quadratic regression of colony size on survival, with the resulting selection gradients ( 30) specifying the form and direction of selection ( 31). The wide range in cliff swallow group size is probably maintained by fluctuating survival selection and represents the first case, to our knowledge, in which fitness advantages of different group sizes regularly oscillate over time in a natural vertebrate population. Averaged across years, there was no net directional change in selection on colony size. Oscillating selection on colony size likely reflected annual differences in food availability and the consequent importance of information transfer, and/or the level of ectoparasitism, with the net benefit of sociality varying under these different conditions. Directional selection was predicted in part by drought conditions: birds in larger colonies tended to be favored in cooler and wetter years, and birds in smaller colonies in hotter and drier years.
Colony size was under both stabilizing and directional selection in different years, and reversals in the sign of directional selection regularly occurred. We used a 30-y mark–recapture study of colonially breeding cliff swallows ( Petrochelidon pyrrhonota) to show that the survival advantages of different colony sizes fluctuated among years. Because individuals in certain sizes of groups often have higher apparent fitness than those in other groups, why wide group size variation persists in most populations remains unexplained. Most animal groups vary extensively in size.