Abstract
Aim To separate the effect of overall resource level from the effect of seasonality on avian clutch size to test A shmole's hypothesis that birds have larger clutch sizes in seasonal environments due to high per capita food availability during the breeding season. Location S outh A frica and L esotho. Methods We used a large‐scale environmental gradient to test the effects of maximum resource availability and resource seasonality (estimated by the normalized difference vegetation index) on clutch size variation among local passerine assemblages (25 km × 25 km grid cells). The importance of maximum resource availability was distinguished from the importance of resource seasonality by using multivariate general additive models and by subsetting the data so that variation in one of these parameters was minimized. Spatial autocorrelation was controlled for by using spatial generalized least squares. Results Assemblage mean clutch size showed a hump‐shaped relationship with maximum resource availability but an increase with resource seasonality. When the variation in maximum resource availability was fixed, clutch size increased with increasing seasonality, but it decreased with increasing maximum resource availability when we fixed the variation in seasonality. These results hold for all feeding guilds except granivores, for which we found opposite patterns. The patterns were much less pronounced when family membership was controlled for, indicating that the overall trends are mostly driven by variation between families. Main conclusions Although clutch size can be affected by many factors related to environmental productivity and its variation, A shmole's hypothesis provides the most parsimonious explanation of the observed patterns: geographical patterns in mean clutch size across bird assemblages seem to be driven by variation of per capita food availability determined by seasonal variation of population density.