How traits at multiple levels of biological organization evolve in a correlated fashion in response to directional selection is poorly understood, but two popular models are the "behavior evolves first" (BEF) and the "morphology - performance - behavior - fitness" (MPBF) paradigm. Both acknowledge that behavior can serve as a pacemaker of evolution because selection often acts directly on behavior (i.e., what an animal does in various situations) and that, when behavior evolves, other traits (e.g., subordinate traits, components of Darwinian fitness) will likely do so with some lag. We studied correlated responses to selection in the High Runner (HR) mouse selection experiment, in which 4 replicate lines have been bred for voluntary wheel running distance on days 5&6 of a 6-day exposure to wheels as young adults, as compared with 4 non-selected Control (C) lines. We analyzed a wide range of traits measured at generations 20-24, coinciding with the point at which all HR lines were reaching selection limits.
Based on the MPBF model, we expected the greatest divergence between the HR and C lines for wheel running and its components (average running speed, duration of daily activity), with less divergence in traits at the level of life history (e.g., litter size) and performance (maximal oxygen consumption during forced exercise [VO2max], maximal sprint speed, rota-rod), and the least amount of divergence for such lower-level traits as organ masses and bone dimensions.
HR and C lines were compared with SAS Procedure Mixed, where linetype (HR vs C) was a fixed effect and replicate line was nested within linetype as a random effect. Degrees of freedom for testing the linetype effect were always 1 and 6. Covariates, such as body mass for organ masses and VO2max, were used as appropriate. All analyses were split by sex, although not all traits were measured for both sexes. We used P values as a measure of effect size and the positive False Discovery Rate (pFDR) to control for multiple comparisons (SAS Procedure MULTTEST).
Data were available for more than 226 traits, and the P value was < 0.05 for 68 of these. Not surprisingly, the 4 traits with the lowest P values were all aspects of wheel running measured on days 5&6. The next lowest P value was for activity in a home cage with no wheel present. P values 6-14 were other aspects of wheel running, including when measured for only 1 day or up to 4 weeks. Other traits among the lowest 27 P values included body mass, the size of the femoral condyle, and baseline (resting) plasma corticosterone levels. Surprisingly, the 3 measures of performance showed no evidence of having diverged between the HR and C lines, nor did any of the life history traits.
Overall, results suggest that the HR lines of mice had evolved primarily by changes in motivation, rather than performance ability, at the time they were reaching selection limits. In addition, neither the BEF nor the MPBF models of hierarchical, correlated evolution provide a good fit to the HR mouse selection experiment.