Rapid and longer-term effects of selective breeding for voluntary exercise behavior on skeletal morphology in house mice

Abstract

Selection experiments can elucidate the varying course of adaptive changes across generations. We examined the appendicular skeleton of house mice from four replicate High Runner (HR) lines bred for physical activity on wheels and four non-selected Control (C) lines. HR mice reached apparent selection limits between generations 17 and 27, running ~3-fold more than C. Studies at generations 11, 16, and 21 found that HR mice had evolved thicker hindlimb bones, heavier feet, and larger articular surface areas of the knee and hip joint. Based on biomechanical theory, any or all of these evolved differences may be beneficial for endurance running. Here, we studied mice from generation 68, plus a limited sample from generation 58, to test whether the skeleton continued to evolve after selection limits were reached. Contrary to our expectations, we found few differences between HR and C mice for these later generations, and some of the differences in bone dimensions identified in earlier generations were no longer statistically significant. We hypothesize that the loss of apparently coadapted lower-level traits reflects (1) deterioration related to a gradual increase in inbreeding and/or (2) additional adaptive changes that replace the functional benefits of some skeletal changes.

Keywords: adaptation; artificial selection; behavior; bone; coadaptation; experimental evolution; locomotion.

FIGURE 1

(a and b) Mean distal Ilium width in relation to body length. For males only, the effect of body length was positive and statistically significant. HR males had significantly narrower distal ilia for a given body length, and mini‐muscle males had thinner distal ilia when compared to normal‐muscled males. For females, there was no signifcant effect of either linetype or mini‐muscle

FIGURE 2

(a and b) Mean pelvis mass in relation to body length. For males only, the effect of body length was positive and statistically significant. HR males had significantly lighter pelvises for a given body length, and mini‐muscle males had lighter pelvises when compared to normal‐muscled males. For females, there was no signifcant effect of linetype, but mini‐muscle mice had lighter pelvises only at larger body lengths

FIGURE 3

(a and b) Mean femoral greater trochanter in relation to body length. HR male, and female mice had significantly thinner femoral greater trochanters for a given body length, and mini‐muscle females had thinner femoral greater trochanters as compared with normal‐muscled females

FIGURE 4

(a and b) Mean humerus mass in relation to body length. For males only, the effect of body length was positive and statistically significant. HR males had significantly lighter humeri for a given body length, and mini‐muscle males had lighter humeri when compared to normal‐muscled males. For females, there was no signifcant effect of either linetype or mini‐muscle

FIGURE 5

(a and b) Mean lower ilium length in relation to body length. For both sexes, the effect of body length was positive and statistically significant. Mini‐muscle males had significantly longer lower ilia for a given body length. For females, there was no signifcant effect of either linetype or mini‐muscle

FIGURE 6

Mean femoral dimensions across generations. Femoral dimensions are plotted across generations separately for males and females. Values are least square means and associated standard error bars, with body mass as a covariate. Age at sampling varied among generations. (a and b: HR male mice had thicker femoral heads when compared with C mice in earlier generations, but these differences were not apparent at generation 69. HR female mice had thicker femoral heads throughout the generations sampled, although these differences were not always statistically significant. (c and d): HR male mice had thicker distal femora when compared with C mice at generation 11, but this difference was not statistically significant at generations 21 or 69. HR female mice had thicker distal femora throughout the generations sampled, but especially so at generation 16

FIGURE 7

Power increased with the magnitude of the simulated difference between HR and C lines or between mini‐muscle and normal individuals. When the magnitude of the HR versus C difference exceeded 6%, the power to detect a linetype effect was slightly higher when in the Mini50 models when compared with the MiniFix models, although the increase in power was never greater than ~0.04. The power to detect a mini‐muscle effect also increased with the magnitude of the effect, but the power showed little difference between Mini50 and MiniFix. Under the null hypothesis, the Type I error rates for the mini‐muscle effects were very close to the expected 5%, but for the HR versus C effect it was greatly deflated (~1.4%)