At each flush phase after a bottleneck, additive genetic variances and covariances for these eight traits were obtained from the ancestral population using midparental-offspring covariances (three daughters per family; approximately 30 families per episode), to obtain an additive genetic variance-covariance matrix, used as a basis for measuring generalized genetic distance, D, between each bottlenecked line and the ancestor (Lande 1979; Bryant and Meffert 1990).
1/2], where t represents transpose, d is a column vector of differences in the means of eight traits between the bottlenecked line and the ancestor, A is a matrix of eigenvectors of the pooled G, and [[Lambda].
The bottlenecked lines rebounded quickly after each bottleneck episode, so t can be directly equated to the number of bottleneck episodes (Nei et al.
The generalized genetic distances of the bottlenecked lines from the ancestor are shown in Figure 1 (and Appendix II), separated for size (major axis) and shape (summarized across axes II, III, and IV) components over the course of the five serial bottlenecks, and the randomized blocks ANOVA is given in Table 1.
Therefore, the apportionment of divergence to shape rather than size was greater for the less severe bottlenecks, suggesting that populations subjected to intermediate bottlenecks may be more evolutionarily responsive than those subjected to extreme bottlenecks, also suggested by the magnitudes of additive genetic variance remaining in these bottlenecked lines (Bryant et al.