1. In many species of birds the mean clutch size is not constant through the breeding season. Some species show a seasonal pattern of declining clutch size, while others can show a mid-season peak. We suggest an evolutionary interpretation of the different patterns of seasonal change in birds, and test predictions by analysis of nest-record data held by the British Trust for Ornithology (BTO). 2. We assume that conditions for breeding will show a seasonal pattern of improvement and then decline, and that the optimal clutch size (Lack 1947) is larger for each individual in the population when the conditions for breeding are better. 3. In multi-brooded species, seasonal reproductive success is determined partly by the number of broods that parents can raise during the breeding season, not simply by the success of a single brood. Thus, we suggest that individuals in multi-brooded species should be selected to start breeding before the date when their optimal clutch size is greatest, while single-brooded species should await the date when the optimal clutch size is greatest. We predict that multi-brooded species will show a seasonal pattern of increase and then decrease in mean clutch size, with a mid-season peak, while the mean clutch size of single brooded species should decline through the season. 4. We calculated four indices of seasonal change, derived from the parameters of quadratic models fitted to the data for 66 species. The data support our predictions by showing that in increasingly multi-brooded taxa: (a) the ratio of the maximum clutch to the clutch size at the beginning of the breeding season increased, (b) the slope of the quadratic curve at the beginning the season increased, (c) the maximum clutch size occurred later in the breeding season, and (d) the modal laying date became earlier with respect to the date of maximum clutch size. Furthermore, (e) early in the breeding season clutch size declined with date in single-brooded taxa, but increased in multi-brooded taxa. The patterns were similar for passerines and non-passerines, but migrant families did not show clear patterns. 5. The multi-brooded blackbird, Turdus merula L., and single-brooded great tit, Parus major L., at single study sites showed similar patterns to that estimated from the BTO data which were pooled over sites and years. Thus pooling data is unlikely to confound the general results. 6. We conclude that multi-brooded species are selected to start laying early, so that the first young hatch before conditions are optimal for raising nestlings, while single-brooded species await better conditions. However, migrants appear to start laying later compared with residents with respect to the date of the peak clutch size. We suggest that migrants suffer constraints which prevent laying until after the date when the optimal clutch size is greatest.
