The effects of a single population bottleneck of differing severity on heritability and additive
genetic variance was investigated experimentally using a butterfly. An outbred laboratory stock
was used to found replicate lines with one pair, three pairs and 10 pairs of adults, as well as
control lines with approximately 75 effective pairs. Heritability and additive genetic variance of
eight wing pattern characters and wing size were estimated using parent–offspring covariances in
the base population and in all daughter lines. Individual morphological characters and principal
components of the nine characters showed a consistent pattern of treatment effects in which
average heritability and additive genetic variance was lower in one pair and three pair lines than in
10 pair and control lines. Observed losses in heritability and additive genetic variance were
significantly greater than predicted by the neutral additive model when calculated with coefficients
of inbreeding estimated from demographic parameters alone. However, use of molecular markers
revealed substantially more inbreeding, generated by increased variance in family size and
background selection. Conservative interpretation of a statistical analysis incorporating this
previously undetected inbreeding led to the conclusion that the response to inbreeding of the
morphological traits studied showed no significant departure from the neutral additive model. This
result is consistent with the evidence for minimal directional dominance for these traits. In
contrast, egg hatching rate in the same experimental lines showed strong inbreeding depression,
increased phenotypic variance and rapid response to selection, highly indicative of an increase in
additive genetic variance due to dominance variance conversion.