The energetics of movement, from exercise to ecology and evolution

Movement is a defining and essential feature of the animal kingdom. Nearly all animals today must walk, run, swim, climb or fly through their environments to forage, find mates and evade predators. The energy cost of locomotion can be considerable, accounting for the large majority of metabolic expenditure while on the move. The high cost of movement places considerable demands on the body and can impact the energy available for vital tasks such as growth, reproduction and maintenance (Pontzer and McGrosky, 2022). Consequently, energy expenditure in locomotion is a major focus in animal physiology, ecology and evolution.

Early work on the energetics of locomotion focused on the laboratory measurements of movement. Those foundational studies established that the rate of energy expenditure during locomotion could easily reach an order of magnitude above resting expenditure, driven by the muscle activity needed to propel the body (Taylor et al, 1982; Schmidt-Nielsen, 1972). As research on animal movement expanded to focus increasingly on wild populations and longer timeframes, the impact of locomotion on daily energy expenditure has proven to be less straightforward. Some ecological studies of daily expenditure have shown a direct correspondence to daily movement, but others have not. The effect of locomotor energy expenditure on the evolutionary changes in species' daily energy expenditure (also called ‘field metabolic rate’) is even less clear.

In this Review, I examine locomotor energy expenditure over different timescales. I begin by discussing the physiological determinants of locomotor cost, focusing on recent work linking the physiology of muscle contraction to metabolic expenditure. I then discuss the impact of locomotor cost over increasingly longer timeframes, from minutes to years, and through evolution (Fig. 1). Over short durations, the cost of locomotion is clearly evident in the elevated rate of energy expenditure. Over longer periods, the energy costs of locomotion may be more apparent in the trade-offs imposed on other tasks. This duration-dependent perspective is useful in integrating laboratory, field and evolutionary investigations of locomotor cost, and provides a framework for incorporating physical activity into broader considerations of ecology and life history (White et al., 2022; Pontzer and McGrosky, 2022; McGrosky and Pontzer, 2023; McNamara and Houston, 2008).

During locomotion, whether running, flying, climbing or swimming, energy expenditure is dominated by muscle contractions. These contractions use the energy liberated when adenosine triphosphate (ATP) is converted to adenosine diphosphate (ADP) to power the actin–myosin cross-bridge cycles that develop muscle tension and run the calcium pumps of the sarcoplasmic reticulum (Barclay and Curtin, 2023). Some of this muscle activity is in the service of mechanical work, for example to lift the body's center of mass or to move the limbs (Fedak et al, 1982; Heglund et al, 1982; Daley and Biewener, 2003; Marsh et al., 2004). Marsh and colleagues (Marsh et al., 2004) demonstrated that the mechanical work needed to swing the legs constitutes roughly 26% of the cost of locomotion in running guinea fowl, for example. During some modes of locomotion, such as climbing up vertical substrates (Hanna et al, 2008) or flapping flight (Tucker, 1966), the mechanical work needed for movement provides an accurate predictor of energy consumption.

Often, however, mechanical work is a poor predictor of locomotor cost, at least during walking and running on level ground (Fedak et al, 1982; Heglund et al, 1982). Studies in the 1990s established that the energy cost of walking and running is strongly predicted from the magnitude and frequency of limb forces applied to the ground to support body weight (Kram and Taylor, 1990; Roberts et al., 1998a,b). As the length of the muscle changes very little, the amount of mechanical work performed is also modest, even as the volume of muscle activated to generate the ground force is substantial. When running or walking on level ground, the integral of ground force produced by the limbs while in contact with the substrate must equal the integral of body weight over the entire stride cycle, and therefore shorter contact times necessitate greater forces (Kram and Taylor, 1990). Shorter contact times require greater ground forces and thus more muscle to be activated. Muscle activation costs are considerable and distinct from the energy consumed in cross-bridge cycling (Barclay and Curtin, 2023), and thus shorter contact times and greater muscle activation incur increased metabolic costs of locomotion separate from those associated with mechanical work.

Ground forces and muscle activation correctly predict variation in the cost of terrestrial locomotion with speed and across species. Walking and running at faster speeds typically reduces contact times, resulting in higher ground forces and greater rates of energy expenditure (Kram and Taylor, 1990). The volume of muscle activated to produce a unit of ground force is broadly independent of body size among legged vertebrates (Biewener, 1989), but those with comparatively poor limb-effective mechanical advantage must activate more muscle and exhibit correspondingly greater cost (Roberts et al., 1998a,b; Pontzer et al, 2009). Within species, operating the muscle at a less favorable sarcomere length or shortening velocity can increase the volume of muscle needed to produce a given force, thereby increasing cost (Beck et al., 2020, 2022). Across species, those with shorter limbs use shorter steps and incur greater forces and costs, resulting in the well-established allometry of locomotor energy cost for walking and running (Taylor et al, 1982; Pontzer, 2007).

More recent work has attempted to integrate work- and force-based models of locomotor cost. One approach models locomotor cost as the sum of muscle activation and work (Pontzer, 2016). This activation, relaxation and cross-bridge (ARC) model accurately predicts the energy cost of walking, running and climbing over a range of inclines, from steep downhill to level ground to vertical climbing, and across range of species, from cockroaches to cows (Pontzer, 2016). Lichtwark and colleagues have demonstrated success predicting locomotor costs using a similar approach within species, integrating activation and work costs in computer models of muscles to accurately predict the energy cost of hopping at different heights and frequencies in humans (Jessup et al., 2023). Notably, within the ARC model, the ratio of work cost to activation-relaxation cost varies allometrically and across locomotor modes. Work costs predominate during climbing but are relatively small during walking and running on level ground, and work accounts for a greater proportion of running cost as body mass increases (Pontzer, 2016).

Nearly all of the work to date investigating the contribution of muscle activation and cross-bridge cycling to locomotor cost have been confined to mammals and birds during walking, running and climbing. Nonetheless, the fundamental physiology of actin-myosin cross-bridge cycling and calcium handling in birds and mammals appear to be broadly shared with fish, insects and other animals (Cao and Jin, 2020), and therefore we can expect activation–relaxation and work costs to contribute in a similar manner across taxa and locomotor modes. For example, the U-shaped cost of transport (energy/distance) with speed is consistent with the change in power requirements to generate lift and thrust, suggesting mechanical work likely dominates the cost of flapping flight (Butler, 2016; Salehipour and Willis, 2013). During gliding, muscle contraction is mostly isometric, and activation and relaxation costs may dominate (Taylor et al, 2016; Duriez et al., 2014). Swimming costs in fishes are attributable to the mechanical work done to generate hydrostatic forces (Lauder and Di Santo, 2015), and the rate of expenditure increases with swim speed and the muscle forces exerted to propel the body (Alfonso et al., 2021; Zhang and Lauder, 2024).

Research linking muscle physiology to locomotor energetics is ongoing, but the general framework emerging over the past several decades appears sound: energy expenditure during locomotion results from muscle contraction (both activation and the cross-bridge cycling) to exert force on the substrate and to do work. The proportion of activation (or force) versus work cost will vary with body size and locomotor mode, but together they can raise whole-body energy expenditure by ten to twenty times above that associated with resting (Taylor et al, 1982; Tucker, 1966; Schmidt-Nielsen, 1972). Over the shortest timescales, during acute locomotor bouts, muscle activity to move the body accounts for the large majority of the body's metabolic energy expenditure (Fig. 1).

Advances in wearable tracking devices have revolutionized the science of locomotor ecology over the past two decades. Many of these units are equipped with accelerometers and provide a recording of the animal's movement that can, in principle, be used as a proxy or predictor of energy expenditure. When these devices are used in tests lasting several minutes to an hour (e.g. treadmill or wind tunnel studies), the magnitude of the accelerometry signal generally corresponds very closely to the rate of energy expenditure measured by oxygen consumption or heart rate during running, flying and swimming (Halsey et al, 2011; Alfonso et al., 2021; Hicks et al., 2017).

Accelerometers measure forces, and so the agreement between accelerometry and energy expenditure is unsurprising given the importance of force production in locomotor cost. It is notable, however, that the relationship between accelerometry and energy expenditure differs between species and among tasks (Halsey et al., 2008, 2011). This variation likely reflects, at least in part, differences in the relative contribution of muscle force versus work during locomotion. In humans, for example, the slope between energy expenditure and overall dynamic body acceleration (ODBA), a common accelerometry metric, is steeper during running than during walking (Halsey et al., 2008). Ladds and colleagues found no relationship between vector dynamic body acceleration (VeDBA), another common acceleration metric similar to ODBA, and energy expenditure during swimming and diving in captive sea lions and fur seals (Ladds et al., 2017). Ladds and others (Ladds et al., 2017; Wilson et al., 2020) have suggested that ODBA may be particularly poor at predicting energy expenditure in marine mammals, as the energy cost of movement may simply reduce the energy expenditure needed to maintain body temperature in cold water, resulting in little or no net change in whole-body expenditure in swimming versus resting. Muscle work accounts for a greater proportion of locomotor cost than force production during swimming, which could also contribute to the discrepancy between accelerometry and energy expenditure.

Dozens of studies have deployed small, wearable accelerometers to track the daily movements of vertebrates in the wild (Wilson et al., 2020). In several studies, accelerometer recordings have been paired with doubly labeled water measurements of daily energy expenditure (also called ‘field metabolic rate’, J day⁻¹), allowing researchers to determine the impact of locomotor cost on a validated and independent objective measure of energy expenditure over multiple days. Much of this work has been done with birds, often during reproductive seasons in which adults are provisioning chicks. Studies in Adélie penguins (Hicks et al., 2020), Australasian gannets (Sutton et al., 2023), pelagic cormorants (Stothart et al., 2016), and murres (Elliott et al., 2013), for example, have reported strong correlations between VeDBA or ODBA and doubly labeled water measures of daily energy expenditure. In mammals, accelerometry has been shown to correlate with doubly labeled water measures of daily expenditure in Northern seals and fur seals (Jeanniard-du-Dot et al., 2017), and in polar bears (Pagano and Williams, 2019).

Results from other studies of daily energy expenditure and accelerometry have been mixed. Ste-Marie and colleagues (Ste-Marie et al., 2022) found that ODBA did not predict doubly labeled water daily energy expenditure in dovekies, for example. Sutton and colleagues (Sutton et al., 2021), in a study of little penguins, found that VeDBA predicted daily expenditure for adults who actively foraged at sea during the measurement period, but not for those who stayed on land. In numerous human studies, accelerometry measures of daily physical activity typically explain ∼20% or less of the variance in doubly labeled water measured daily energy expenditure (Plasqui et al, 2013; Pontzer et al., 2016).

Implantable heart rate loggers have provided another method for estimating energy expenditure in the field (Halsey et al., 2008, 2019; Hicks et al., 2017) and may shed additional light on the discrepancy between daily movement and expenditure. As in treadmill studies, heart rate generally increases with ODBA and VeDBA, although the slope of the relationship often varies among activities (Hicks et al., 2017). Heart rate monitoring has also highlighted the impact of other, non-locomotor factors on energy expenditure. For example, in graylag geese, average daily heart rate was shown to increase during the incubation period in females (but not males) and to vary seasonally with environmental temperature (Wascher et al, 2018). Heart rate also responds to acute psychological stressors (Wascher, 2021; Sawai et al., 2007; Seematter et al., 2000, 2002). Variation in these non-locomotor tasks can obscure the relationship between physical activity and daily energy expenditure.

Wilson and others have discussed various reasons that accelerometry might fail to predict variation in daily expenditure (Wilson et al., 2020). When animals are relatively sedentary and energy expenditure is dominated by non-locomotor tasks, such as thermoregulation or digestion, accelerometry is likely to be a poor predictor of expenditure. Human studies certainly support this view. For athletes during competition, daily expenditure corresponds strongly with activity workload (Thurber et al., 2019; Cooper et al., 2011; Plasqui et al., 2019; Best et al., 2023), but accelerometry fares much more poorly at the lower levels of physical activity typical of everyday life (Plasqui et al, 2013; Pontzer et al., 2016). The impact of locomotion on energy expenditure over timescales of hours to days, then, is dependent on the degree of physical activity (Fig. 1).

As the timescale of analysis increases to weeks, months and years, the relationship between locomotor activity and daily energy expenditure becomes increasingly tenuous. Animals appear to acclimate to long-term changes in habitual physical activity in ways that maintain daily energy expenditure within a narrow window (Halsey et al., 2019; Pontzer, 2015a). Consequently, over these timescales, daily expenditure often shows little correspondence to locomotor activity.

This phenomenon of metabolic acclimatization, sometimes called ‘constrained energy expenditure’ (Pontzer, 2015a, 2018), has been demonstrated in experimental studies across a range of species. When food availability is manipulated to increase the work needed to forage, species from mice to zebra finches to hummingbirds have been shown to increase daily locomotor activity without corresponding increases in daily expenditure (Pontzer, 2015a; Wiersma and Verhulst, 2005; Deerenberg et al., 1998; Perrigo, 1987; Perrigo and Bronson, 1983). More recently, O'Neal and colleagues manipulated wheel access for laboratory mice without affecting food access (which was always ad libitum) and reported no change in daily expenditure over 3 weeks even as daily wheel running steadily increased (O'Neal et al., 2017). Similar results are seen in human exercise studies. In exercise interventions lasting ∼6 months or longer, the increase in daily energy expenditure from pre-intervention to the end of the study is typically less than the prescribed exercise dose (Broskey et al., 2021; Donnelly et al., 2003; Willis et al., 2014; Goran and Poehlman, 1992; Flanagan et al., 2024). In one 16-month study, participants were assigned 2000 kcal per week of supervised exercise, but their daily energy expenditures during the final weeks of the study were not statistically significantly different from their sedentary, pre-intervention baseline (Donnelly et al., 2003).

Measuring the effects of physical activity on energy expenditure over several months or years presents logistical challenges for experimental design. One approach is to compare populations that differ in habitual locomotor activity and have therefore lived with different locomotor demands for years or even decades. My colleagues and I have measured daily energy expenditures in human populations with high levels of daily physical activity, including hunter-gatherers, farmers and pastoralists. Despite engaging in far more physical activity each day, people in these subsistence populations have similar daily energy expenditures to comparatively sedentary people in the USA and other industrialized populations, in analyses that control for body size and fat percentage (McGrosky et al., 2024; Urlacher et al., 2021; Pontzer et al., 2012; Gurven et al., 2016). Similarly, in regression analyses accounting for body size, daily energy expenditures for non-human primates in captivity do not differ from those in the wild (Pontzer et al., 2014). In pandas, sheep and kangaroos, reported daily expenditures for populations housed in captivity do not differ from free-ranging groups (Munn et al., 2013; Nie et al., 2015).

Constancy in daily energy expenditure despite substantial variation in physical activity suggests trade-offs, whereby greater expenditure on locomotion is offset by reduced expenditure in other physiological tasks. Wilson and colleagues discuss the potential for thermoregulatory trade-offs, in which locomotion generates heat that reduces the need to burn energy to maintain body temperature (Wilson et al., 2020). This movement–heat trade-off has been noted by others as well (Ocobock, 2016; O'Neal et al., 2017).

The metabolic trade-offs imposed by increased physical activity appear to be much broader than thermoregulation, however, affecting all aspects of physiology, including growth, maintenance and reproduction (Pontzer and McGrosky, 2022). For example, adolescent mice grew more slowly and were less likely to achieve ovulation when they had to run further each day (Perrigo and Bronson, 1983). In adult house mice and deer mice, increased wheel running resulted in smaller or fewer pups (Perrigo, 1987). For zebra finches, increased foraging effort led to slower feather regrowth and an increased interval between clutches (Wiersma and Verhulst, 2005). In humans, people who are more physically active have lower chronic inflammation and white blood cell counts (Klasson et al, 2022; Pontzer, 2018), indicating reduced investment in immune function. In physically active subsistence populations, providing wells to alleviate the physical burden on women to gather water has been shown to increase fertility (Gibson and Mace, 2006). Conversely, in comparatively sedentary industrialized populations, endurance exercise lowers reproductive hormone levels (Jasieńska and Thune, 2001; Hackney and Lane, 2018; Pontzer, 2018). These responses to physical activity are likely an important reason that exercise is generally beneficial for health, and why, conversely, extreme exercise workload in athletes can lead to the clinical suppression of the immune system and reproductive system seen in overtraining syndrome or relative energy deficiency (Pontzer, 2018; Stellingwerff et al., 2021).

The high levels of daily expenditure observed during periods of high-intensity activity, whether provisioning chicks at the nest (Drent and Daan, 1980; Elliott et al., 2013; Stothart et al., 2016; Sutton et al., 2023), nursing pups (Speakman, 2008; Speakman and Król, 2010), migrating (Brown et al., 2023; Gill et al., 2009; Hedenström, 2010; Wikelski et al., 2003) or during athletic competition (Plasqui et al., 2019; Thurber et al., 2019), would seem to conflict with the observation that daily expenditures are constrained. Thurber and colleagues (Thurber et al., 2019), using data from human studies, provide a framework for reconciling these seemingly contradictory findings. Analyzing data from extreme athletic events, they showed that the maximum daily energy expenditure is strongly related to event duration: the longer the event, the lower the maximum sustained daily expenditure (Thurber et al., 2019). Expressing daily expenditure as metabolic scope (i.e. multiples of basal metabolic rate), humans were able to maintain a metabolic scope of ∼8.5 for a 1-day ultramarathon, ∼4.9 for the 22-day Tour de France, and ∼3.5 for a 70-day Arctic trek. As duration approached 1 year, the maximum sustainable expenditure for humans was a metabolic scope of ∼2.5. To date, all human populations measured during normal daily life (i.e. not during athletic training or competition) have average metabolic scopes of below 2.5 regardless of lifestyle, and the large majority are below 2.0 (Dugas et al., 2011; Pontzer, 2015b). The potential impact of locomotor activity on daily expenditure is large over periods of days or weeks, but is constrained over longer timescales.

Data to test this model of metabolic management in other species are scarce, because work on metabolic ceilings in non-humans has focused on identifying a single, maximum sustainable rate of expenditure rather than considering the effect of event duration (Speakman and Król, 2010; Hammond and Diamond, 1997; Peterson et al, 1990). The evidence from humans, as well as the laboratory experiments discussed above in which daily activity levels are manipulated in non-humans, suggests that compensatory mechanisms can act to maintain long-term daily expenditures well below species' short-term maximal expenditure.

Multi-month monitoring of heart rate in wild populations of mammals, birds and fish supports the view that compensatory mechanisms to reduce non-locomotor expenditures when daily activity levels increase are widespread among vertebrate species (Halsey et al., 2019). Two case studies, tracking flight behavior and modeling daily expenditures in wild bird populations, provide further support for this view. In lesser black-backed gulls, a migrating seabird in the eastern Atlantic (Brown et al., 2023), individuals in the study population follow different migration strategies over the year, with some traveling less than 250 km and others migrating over 4500 km. Daily energy expenditures during peak flying periods were twice as high in the long-distance migrators, but, over the full year, average expenditures were more moderate and did not differ from short-distance migrators (Brown et al., 2023). In a 6-month study of common eiders, the average time spent flying varied nearly threefold, between ∼5 and ∼14 min day⁻¹ among individuals (Pelletier et al., 2008). However, despite the high cost of flight (i.e. J min⁻¹) in this species, average daily heart rate (a proxy for daily energy expenditure) did not correlate with time spent flying (Pelletier et al., 2008). Instead, flight time was negatively associated with average heart rate during other parts of the day, indicating compensation to daily flight costs that kept total daily expenditures in check (Pelletier et al., 2008).

Work is needed to understand the timing and degree of energy constraint across species and the underlying physiological and behavioral mechanisms. What is clear with the data available is that the relationship between physical activity and daily expenditure is dependent upon the timescale of analysis. Daily energy expenditure over the course of hours or days can be high and variable, dominated by locomotor expenditure. Over longer timescales of weeks to years, compensatory responses to physical activity and constraints on sustained expenditure serve to moderate daily expenditure, such that the long-term average is lower, less variable, and less dependent on locomotor activity (Fig. 1).

Larger species burn more energy each day, but there is substantial interspecific variability in total daily energy expenditure (i.e. field metabolic rate) when controlling for the effects of allometry (McGrosky and Pontzer, 2023). Data needed to test the relationship between daily expenditure and locomotor ecology are scarce, but the available evidence indicates that evolved increases in daily physical activity do not result in correspondingly higher daily energy expenditures.

Terrestrial mammalian carnivores provide one example of the discordance between locomotor ecology and field metabolic rate. Daily travel distance tends to increase with body mass among terrestrial mammals, but trophic level has a clear effect as well: for a given body size, carnivores travel roughly three to four times farther than herbivores each day (Garland, 1983; Carbone et al., 2005). There is no evidence for reduced cost of transport (i.e. energy/distance) among terrestrial mammals (Taylor et al, 1982), which indicates that the cost of their longer daily travel distances is not offset by greater economy of movement. Yet, compared to other mammalian groups, carnivores do not have elevated daily energy expenditures (McGrosky and Pontzer, 2023; Nagy et al, 1999), suggesting that their greater daily travel costs are offset or otherwise obscured by variation in energy expended on other tasks. Primates provide another point of comparison. DeCasien and colleagues (DeCasien et al., 2024) examined daily energy expenditures among primate species and found no effect of home range (n=37 species) or daily travel distance (n=30 species) in analyses controlling for body size and phylogenetic relatedness.

The lack of correspondence between locomotor ecology and daily energy expenditure could be due in part to the modest proportion of the energy budget that goes to locomotion in many species. Garland, in the foundational paper on this topic, estimated that the percentage of daily expenditure attributed to daily travel (which he termed the ‘ecological cost of transport’) was generally less than 10% for terrestrial mammals (Garland, 1983). More recent estimates, drawing on much larger datasets for field metabolic rate and daily travel distance, suggest most terrestrial mammals expend less than 5% of their daily expenditure on locomotion (Pontzer, 2012). Pelletier and colleagues (Pelletier et al., 2008) give a similar estimate, 4–5%, for average daily flight costs in common eider ducks. Daily locomotor costs may represent a substantially larger portion of daily cost in some species, particularly those that spend more time in flapping flight or other costly forms of locomotion. Brown and colleagues, in their study of lesser black backed gulls, estimated that ∼30% of daily energy expenditure, averaged across the year, was attributable to flight (Brown et al., 2023). Nagy and colleagues, in an early study using data loggers and doubly labeled water to integrate locomotor ecology and energetics, estimated that jackass penguins expend roughly 30% of their daily energy swimming (Nagy et al., 1984). Yet even with these larger estimates, locomotor costs would account for less than half of daily expenditure, and thus variation in other tasks, such as reproduction, growth, immune function, and thermoregulation, could easily obscure the impact of locomotion (Fig. 1).

The discrepancy between daily locomotor costs and daily energy expenditure among species does not diminish the potential importance of locomotor costs in evolutionary ecology or as a target of natural selection. Adaptations to reduce locomotor costs are evident throughout the animal kingdom, from the soaring adaptations in gulls and condors to the streamlining of some fish and other aquatic species, to the use of tendons to store and return energy in running vertebrates. Behavioral adaptations to reduce travel costs by traveling in formation, schooling, or choosing low-cost routes are also well-documented in some species (Owen-Smith et al, 2010; Pontzer, 2020; Green et al., 2020; Zhang and Lauder, 2024; Weimerskirch et al., 2001), highlighting the impact of locomotor cost as a selection pressure on cognitive abilities. Rather than the determinative factor underlying variation in species' daily energy expenditures, locomotor behavior and physical activity should be viewed as one of several, crucial physiological tasks that must be managed within the energy budget, similar maintenance, growth and reproduction (McNamara and Houston, 2008; McGrosky and Pontzer, 2023; Pontzer and McGrosky, 2022). In this light, the daily cost of locomotion might be most usefully measured by its impacts on other tasks and the trade-offs incurred rather than simply through its covariation with daily energy expenditure. Taking this approach, we can integrate locomotor mechanics and cost into the broader framework of evolutionary ecology and physiology.

The interplay of laboratory and field studies has led to important insights in locomotor energetics and biomechanics. Wind tunnel testing of flight costs, combined with accelerometry and global positioning system tracking in wild birds, has deepened our understanding of migration and the pressures it places on energy storage and metabolism (Wikelski et al., 2003; Hedenström, 2010; Brown et al., 2023; Gill et al., 2009). Treadmill studies of walking, running and climbing have been crucial in uncovering the link between locomotor mechanics and muscle energy consumption, and have allowed researchers to reconstruct the energy budgets of wild populations (Pontzer and Wrangham, 2004; Jessup et al., 2023; Kram and Taylor, 1990; Roberts et al., 1998a,b; Pontzer, 2016). Laboratory studies of swimming costs are advancing our understanding of schooling and other ecologically important behavior (Zhang and Lauder, 2024).

The success and utility of this integrative work is clear, but we need to be thoughtful in extrapolating findings from the lab to the field. Many of the salient caveats in applying lab data to field estimates of daily expenditure have been discussed previously (Wilson et al., 2020). This Review suggests an additional element that is often missing from laboratory experiments: time. Experiments take place over minutes or hours, whereas the relevant timeframes in the wild, for animals balancing the demands of foraging, survival and reproduction, might be days, months or years.

Muscle metabolism, through activation and mechanical work, dominate an animal's energy expenditure during locomotion, and can be the primary determinant of daily expenditure over several days of intense activity. Over longer timescales, compensatory changes moderate the metabolic impact of locomotion, such that daily expenditure does not correspond with daily activity. Crucially, these changes appear to impact growth, reproduction and maintenance, incurring real costs to evolutionary fitness (Pontzer and McGrosky, 2022; McGrosky and Pontzer, 2023; McNamara and Houston, 2008). These compensatory responses to long-term changes in physical activity should be considered alongside the acute metabolic costs of locomotion, and they warrant future investigation. Over a lifetime or evolutionary timescales, the cost of locomotion might be better measured in maintenance deferred or fertility diminished rather than just the calories expended to move. A holistic understanding of the cost of locomotion must incorporate its impacts across other physiological systems.

I thank the organizers of this workshop for the invitation to participate, and all attendees for their helpful discussion, insights and collegiality. I thank Andrew Biewener for his groundbreaking research in animal biomechanics and his mentorship, both of which have shaped my own work.