Sleep deprivation negatively impacts both physical and psychological health in both humans and animal models. Exercise, on the other hand, can have beneficial effects on various aspects of physical and mental health. However, little is known about the ways in which sleep deprivation and exercise may interact, especially for exceptionally high levels of exercise. We studied High Runner (HR) mice from a long-term artificial selection experiment to investigate how genetically high exercise level could impact the response to sleep deprivation. A total of 192 adult mice from four replicate HR and four non-selected Control lines (balanced for sex) completed six days of baseline wheel access, followed by three days with or without 6 h/day of total sleep deprivation (TSD) via gentle handling. As expected, HR mice ran farther and faster compared to Controls during days 1-6. TSD reduced the running distance and duration in mice from Control lines, while HR increased running speed and maintained distance (treatment × linetype interaction). TSD-induced changes in body mass differed between linetypes (treatment × linetype interaction): Controls tended to gain mass, whereas HRs lost mass. During the three days prior to TSD, HR mice consistently exhibited more active and fewer maintenance behaviors than Controls. TSD increased resting and decreased wheel activity in Controls but not HRs (treatment × linetype effects significant for both categories). These results demonstrate that genetically based high voluntary activity levels are associated with altered responses to TSD.

Skeletal morphology is determined by a combination of genetic background and phenotypic plasticity induced by mechanical loading across the lifespan (e.g., exercise), among other factors. Changes in limb morphology associated with locomotor activity experienced throughout individual ontogeny and in species over evolutionary time can provide insight into the adaptation of vertebrate locomotor systems. Here, a mouse line artificially selected for increased voluntary wheel-running distance is used to explore the effects of selective breeding on the genetic "baseline" morphology of the hind limb skeleton and to describe its plastic response to chronic exercise, including potential genotype by environment interactions. This experimental design additionally allowed for testing two hypotheses: (1) that bone plasticity itself is a trait that can evolve in direction and magnitude, and (2) that bone plasticity can influence the trajectory of selection for locomotor phenotypes. Using a combination of 3D shape analyses, we find that 82 generations of selective breeding result in substantial changes in the morphology of the hind limb skeleton, as well as subtle changes in the skeletal response to exercise that support the evolvability of bone plasticity (i.e., skeletal plasticity differs based on genetic background). We additionally find that the evolved bone shape of selected mice does not resemble the plastic exercise response of control-line animals. These results provide further evidence of genetic variations among populations in the plastic bone response to mechanical loading and inform our understanding of the plastic and evolutionary lability of the skeletal system in response to locomotor demands.

The capability for sustained running has convergently evolved multiple times in mammals, and involves myriad anatomical, physiological, and behavioral adaptations. The ribcage plays a critical role in both respiration and locomotion but its adaptations to running are largely unexplored. Robustly testing adaptation in wild populations is challenging, so we use artificial selection for voluntary wheel-running behavior (i.e., High Runner or HR mice) to directly test form-function relationships associated with sustained running. We compared ribcage configuration and shape of HR (males: 52, females: 47) to control (males: 48, females: 48) mice using rib counts and 3D Geometric Morphometrics. Two of four HR lines had an additional rib and increased variation in the proportion of true to false ribs, suggesting that ribcage patterning has been impacted by selection. This variability among lines suggests that selection for wheel running has resulted in adaptations that are expressed variably among the selected lines, resulting in "multiple solutions" to selection. Total ribcage shape did not vary significantly between HR and Control mice. Instead, the effect of selection varied along the ribcage, with significant effects in the caudal ribs. Further, the caudal ribcage of HR mice showed increased disparity, within-rib, and among-rib integration compared to controls. The strong response of caudal ribs indicates a modular pattern of adaptation, with cranial ribs possibly constrained by their role in ventilation. This study demonstrates that adaptation in the mammalian ribcage is likely shaped by a complex interplay of selection and craniocaudal integration and may result in variation at multiple anatomical levels (count, shape, modularity).

We studied lifespan and lifetime reproductive success (LRS) in mice from a selection experiment that includes 4 replicate High Runner (HR) lines bred for voluntary wheel-running behavior and 4 non-selected Control (C) lines. HR mice typically run threefold farther on a daily basis, leading to substantially increased energy expenditure and food consumption, and are also more active and eat more when housed without wheels. HR and C mice have been shown to differ for many other traits that could affect reproduction, including body mass and composition, aerobic capacity, hormonal profile (including corticosterone and leptin), and other behavioral traits (e.g., maternal care, thermoregulatory nesting, anxiety-like behavior in an elevated plus maze). At generation 97, male–female pairs were maintained until end of life in standard cages without wheels, with ad lib food and water. The number of pups weaned (LRS) ranged from 0–80, with a mean of 29.1 ± 8.1 (SE) for C lines and 25.8 ± 4.4 for HR (P = 0.73). Thus, artificial selection for high voluntary exercise did not significantly alter LRS. In addition, we found no statistical difference in average lifespan for dams or sires. However, replicate lines differed significantly for several traits, presumably due primarily to random genetic drift. At the level of individual variation, correlations among reproductive traits and lifespan provided little evidence for negative associations, and LRS was not negatively related to parental longevity. Overall, long-term selection on voluntary locomotor behavior did not produce detectable reductions in reproductive output or lifespan under standard laboratory conditions.

Exercise increases muscle and bone strength and mass, but effects on tendons are less documented. We investigated the impact of voluntary exercise (wheel running) during early-life exercise (weanling; 3 weeks old) compared to post-skeletal maturity (young adult; 9 weeks old) on tendon morphology and material properties. We utilized a selectively bred High Runner (HR, N = 40) mouse line and a control line (N = 40). Mice underwent 8 weeks in cages either with or without wheels. HR mice ran ~3-fold more and were smaller than controls, but exercise reduced body mass in both lines. Tendon cross-sectional area was unaffected, but tendon length showed a line*exercise interaction (p = 0.0410) and a near-significant line*age interaction (p = 0.0866). HR mice broadly had greater yield stress (p = 0.0262) and tended toward higher failure stress (p = 0.0676) than controls. Work to failure was greater in younger cohort mice (p = 0.0435), and marginal age-related interactions were observed for modulus (line*exercise, p = 0.0632) and yield strain (line*age, p = 0.0535). HR mice were more responsive to exercise; older exercised HR mice had shorter tendons (p = 0.0282), and younger exercised HR mice showed lower yield and failure strains than sedentary counterparts (p = 0.0445, 0.0246). Exercise and its relative timing produced slight but complex effects on tendon properties, with HR mice showing the strongest structural and mechanical responses.

Various studies have demonstrated adverse effects of dehydration on human athletic and cognitive performance, but most are limited in scope. Moreover, few have studied such effects on voluntary exercise. The purpose of this study was to evaluate the effects of 24 h of water deprivation on voluntary wheel-running behavior (a model for human voluntary exercise) and body mass in laboratory house mice. To increase the probability of detecting the effects of water deprivation, we studied mice from four replicate High Runner (HR) lines that have been bred for high levels of wheel running. Both sexes of HR mice run ∼3-fold more revolutions/day than the four non-selected Control (CON) lines, have increased motivation for wheel running and an increased capacity for aerobic exercise (i.e., ability), and will run voluntarily at speeds nearer to their aerobic capacity. Adults of both sexes were given 6 days of wheel access, as in the routine selective breeding protocol. At the end of day 6, water bottles were removed for ½ of the mice and a 7th day of wheel access was allowed. As expected, body mass significantly decreased in all groups that experienced 24 h without water. As compared with days 5 and 6, wheel-running distance did not significantly change in mice from CON or HR lines with ad lib water, nor did it change in the water-deprived CON mice. However, water-deprived HR mice of both sexes significantly increased the number of revolutions run over 24 h. Given that wheel running is voluntary, presumably because it provides a rewarding experience, we speculate that the significant increase in running after water removal in HR mice may be attributable to “reward substitution,” in which mice replace the reward provided by drinking water (a primary drive) with the rewards derived from running (which may also be a biological drive). Alternatively, mice may have instinctually been attempting to find water, but only those from HR lines have sufficient physical abilities to increase running in spite of adverse effects of dehydration.

A series of terminological, technical, conceptual, and statistical challenges present themselves when trying to study correlations between measures of performance abilities (what an animal can do) and behavioural traits (what an animal chooses to do). We attempt to synthesise literature on this topic, with a specific focus on locomotor performance and behavioural traits measured with standardised tests. We argue that measures of forced performance (e.g. endurance on a motorised treadmill) and voluntary behaviour (e.g. wheel running) often fall along a continuum, sometimes grading into each other. On the performance end of the continuum, tests should measure what an animal can do when motivation is maximal and/or it is given no choice but to exert itself maximally. On the behavioural end of the continuum, tests should capture what animals choose to do of their own free volition, with no experimental attempt to affect motivation. Hence, performance tests attempt to eliminate variation in motivation by forcing all individuals to be maximally motivated, whereas variation in motivation is an inherent component of all behavioural tests. In some cases, however, differentiating between measures of performance versus behaviour can seem almost arbitrary. Moreover, individuals may consistently differ in how willing they are to 'perform' even when 'forced' to do so. We compiled studies reporting any association (covariation, correlation or linear regression) between putative measures of locomotor performance and behaviour in various taxa. The vast majority of those studies report phenotypic correlations, and only a handful have reported genetic correlations or explored potential correlated responses to selection on performance or behaviour. To our knowledge, this is the first global overview of how locomotor performance and behaviour covary in animals, and we believe that our synthesis will be useful to guide future research on locomotor performance and its relationship with other ecologically relevant traits. For example, we argue that a multi-level (co)variance partitioning approach is necessary to gain insights into the importance of how motivation differs across levels (e.g. among- versus within-individual variation, genetic versus environmental variation). Finally, we outline a multitude of compensation and co-specialisation mechanisms that may occur between performance and behaviour, and propose future avenues for research that include selection and manipulative studies to help identify the role of correlational selection, individual experience, and predation detectability on the relationships between behaviour and performance.

In various organisms, sequencing of selectively bred lines at apparent selection limits has demonstrated that genetic variation can remain at many loci, implying that evolution at the genetic level may continue even if the population mean phenotype remains constant. We compared selection signatures at generations 22 and 61 of the “High Runner” mouse experiment, which includes 4 replicate lines bred for voluntary wheel-running behavior (HR) and 4 non-selected control (C) lines. Previously, we reported multiple regions of differentiation between the HR and C lines, based on whole-genome sequence data for 10 mice from each line at generation 61, which was >31 generations after selection limits had been reached in all HR lines. Here, we analyzed pooled sequencing data from ~20 mice for each of the 8 lines at generation 22, around when HR lines were reaching limits. Differentiation analyses of allele frequencies at ~4.4 million SNP loci used the regularized T-test and detected 258 differentiated regions with FDR = 0.01. Comparable analyses involving pooling generation 61 individual mouse genotypes into allele frequencies by line produced only 11 such regions, with almost no overlap among the largest and most statistically significant peaks between the two generations. These results implicate a sort of “genetic churn” that continues at loci relevant for running. Simulations indicate that loss of statistical power due to random genetic drift and sampling error are insufficient to explain the differences in selection signatures. The 13 differentiated regions at generation 22 with strict culling measures include 79 genes related to a wide variety of functions. Gene ontology identified pathways related to olfaction and vomeronasal pathways as being overrepresented, consistent with generation 61 analyses, despite those specific regions differing between generations. Genes Dspp and Rbm24 are also identified as potentially explaining known bone and skeletal muscle differences, respectively, between the linetypes.

Selection experiments play an increasingly important role in comparative and evolutionary physiology. However, selection experiments can be limited by relatively low statistical power, in part because replicate line is the experimental unit for analyses of direct or correlated responses (rather than number of individuals measured). One way to increase the ability to detect correlated responses is through a meta-analysis of studies for a given trait across multiple generations. To demonstrate this, we applied meta-analytic techniques to two traits (body mass and heart ventricle mass, with body mass as a covariate) from a long-term artificial selection experiment for high voluntary wheel-running behavior. In this experiment, all four replicate High Runner (HR) lines reached apparent selection limits around generations 17-27, running approximately 2.5- to 3-fold more revolutions per day than the four non-selected Control (C) lines. Although both traits would also be expected to change in HR lines (relative heart size expected to increase, expected direction for body mass is less clear), their statistical significance has varied, despite repeated measurements. We compiled information from 33 unique studies and calculated a measure of effect size (Pearson's R). Our results indicate that, despite a lack of statistical significance in most generations, HR mice have evolved larger hearts and smaller bodies relative to controls. Moreover, plateaus in effect sizes for both traits coincide with the generational range during which the selection limit for wheel-running behavior was reached. Finally, since reaching the selection limit, absolute effect sizes for body mass and heart ventricle mass have become smaller (i.e. closer to 0).

A putative male advantage in wayfinding ability is the most widely documented sex difference in human cognition and has also been observed in other animals. The common interpretation, the sex-specific adaptation hypothesis, posits that this male advantage evolved as an adaptive response to sex differences in home range size. A previous study a decade ago tested this hypothesis by comparing sex differences in home range size and spatial ability among 11 species and found no relationship. However, the study was limited by the small sample size, the lack of species with a larger female home range and the lack of non-Western human data. The present study represents an update that addresses all of these limitations, including data from 10 more species and from human subsistence cultures. Consistent with the previous result, we found little evidence that sex differences in spatial navigation and home range size are related. We conclude that sex differences in spatial ability are more likely due to experiential factors and/or unselected biological side effects, rather than functional outcomes of natural selection.