Breaking through artificial selection limits of an adaptive behavior in mice and the consequences for correlated responses

Previous divergent selection for nest-building behavior at 22 +/- 1 degrees C resulted in a 40-fold difference between the high and the low lines in amount of cotton used to build a nest. Correlated responses to selection indicated positive genetic correlations with body weight, nest-building at 4 +/- 1 degrees C, and litter size and negative genetic correlations with food consumption. At generation 46, the replicate high-selected (High 1 x High 2), randomly bred control (Control 1 x Control 2), and low-selected (Low 1 x Low 2) lines were crossed and the F1 showed significant heterosis for nest-building behavior. Regression of the F3 on the F2 generation gave heritability estimates of 0.16 +/- 0.10 for the high and 0.07 +/- 0.10 for the low cross, revealing a potential to break the selection limit (at least in the high direction), which had been reached at about 20 generations of selection. Indeed, renewed selection resulted in responses in both the high and the low directions of nesting, yielding realized heritabilities of 0.29 +/- 0.02 and 0.30 +/- 0.004, respectively. Replicated renewed selection, using the F3 generation as the base population, in the high direction of nesting resulted in correlated increases in nest-building at 4 +/- 1 degrees C, litter size, and food consumption. Body weight did not change. The positive correlation with food consumption is opposite in sign compared to the original selection experiment. This indicates that the evolutionary potential of a population to adapt to a changing environment not only depends on its current genetic variability in one adaptive trait, but may be constrained by genetic correlations changing over the course of selection.