Selection for maternal behavior in mice—direct and correlated responses

Abstract

Three generations of divergent selection for a maternal care index (MCI) in mice were practiced in each of three replicates. MCI was recorded for 29–32 dams and litters in each replicate/line subclass. Pups of the top (or bottom) 6–10 dams in each replicate, depending on number of females in the litters, were selected in high (H) and low (L) MCI lines. At birth, litters with more than 10 pups were standardized to 10 pups whereas litters with no more than 10 pups remained intact. On day 4, 8, 12, and 16 of lactation, time budgets during a 30-min session were recorded by real-time video camera. The times spent in the activities of nursing pups, licking pups, retrieving pups, resting with pups, and building a nest were summed, and MCI was the percentage of the total time devoted to these activities. Number born alive (NBA), number weaned (NW), litter weaning weight (LWW), and mating weight (MW) were recorded. The difference between H and L in cumulative selection differential (CSD) of MCI was 59.8%. Regressions of differences in H and L on generation and on cumulative selection differential were 4.878±0.414 and 0.239±0.011 (realized heritability), respectively ( P<0.01). Means for MCI in Generation 3 for H and L, respectively, were 80.7 and 63.7% ( P<0.01). Variances and covariances among traits were estimated with multiple-trait, derivative-free, restricted maximum likelihood procedures. Estimates of heritabilities of MCI, NBA, NW, LWW, and MW were 0.20, 0.31, 0.12, 0.39, and 0.49, respectively. Genetic correlations of MCI with most traits were low; however, a correlation of 0.78 with number weaned was found indicating selection for MCI may increase pre-weaning survival rate. Selection for MCI was effective because it was heritable (0.24) and highly variable (CV=24.8%), but changes in MCI were not strongly associated with litter size at birth, litter weaning weight, or dam mating weight.