Activity, heart rate, and energy expenditure of a cold-climate mesocarnivore, the Canada lynx (Lynx canadensis)

Predators occupying high trophic positions tend to have high energy requirements due to their large body size (Kleiber 1932; Schmidt-Nielsen 1984; Peters 1986; White et al. 2019); elevated resting metabolic rates due to high muscle mass and large organs (Muñoz-Garcia and Williams 2005; but see White and Kearney 2013); and high levels of activity required to locate, pursue, and capture mobile prey (Gittleman and Harvey 1982; Carbone et al. 2005). At the same time, their energetic returns can be highly variable due to their reliance on difficult-to-find, mobile prey, requiring energetic strategies that minimize costs of activity, maximize fasting endurance, and (or) maximize energetic returns.

Within the mammalian order Carnivora, Carbone et al. (1999, 2007) describe two broad categories of predators — small carnivores feeding on small-bodied prey and large carnivores feeding on large-bodied prey — separated at an energetically defined body size threshold of 14.5 kg (see Fig. 1). Below this body size threshold, a combination of the smaller body size of the predator — which reduces overall energy expenditure — and the small body size of the prey — which are abundant, require shorter search distances, and brief capture and killing phases (Griffiths 1980) — makes specialization on smaller bodied prey energetically advantageous and sufficient to meet energetic needs. Above this body size threshold, a high-cost lifestyle, brought on by larger body size, longer search times for prey, faster chases, and greater power requirements, is compensated by greater energetic returns from large-bodied prey, making specialization on large-bodied prey the most energetically feasible strategy. What is less clear, is what strategy makes the most sense for intermediate-sized carnivores (∼10–15 kg, also called mesocarnivores; Roemer et al. 2009; Ray 2000), and particularly, those approaching the 14.5 kg size threshold (Carbone et al. 2007), since they are likely too big to be an energetically efficient small carnivore but too small to be an effective large carnivore that hunts and overpowers large prey.

Insight into how predators of all sizes balance energetic costs and gains across ecologically relevant time and spatial scales has been limited by the logistical challenge of obtaining fine-scale behavioural and metabolic data from free-ranging predators. Although energetic traits of free-ranging carnivores have been estimated previously, they are most often based on activity-specific metabolic rates measured in captivity or predicted allometrically (e.g., Aldama et al. 1991; Young et al. 2012; Newbury and Hodges 2019), or through doubly labeled water over short intervals, without accompanying activity measures (e.g., Winstanley et al. 2003; Dekar et al. 2010; Barbour et al. 2019). These methodological constraints inevitably restrict the types and amount of environmental variation animals are exposed to, the timeframes over which data are collected, and insight regarding multifactorial contributors to variation in energetic traits. Particularly, the inability to track metabolic and behavioural variation of individuals over time has limited in-depth analyses of intraspecific variation in energetic traits, which is not well predicted from interspecific patterns and, in most cases, remains poorly documented and largely unexplained.

However, recent technical, methodological, and conceptual advances are enabling new insight into energetics of free-ranging wildlife via biologged energetic proxies such as acceleration and heart rate. Accelerometers provide both fine-scale activity and behavioural data, providing an effective method to assess energy expenditure based on detailed activity time budgets (Wilson et al. 2006; Halsey et al. 2009; Gleiss et al. 2011; Wilmers et al. 2017). Biologgers recording heart rate have been used in conjunction with accelerometers to provide more complete estimates of energy expenditure across behavioural states, including when animals are at rest (Green et al. 2009; Green 2011). A handful of studies have used accelerometers and GPS collars alongside traditional measures of energy expenditure (i.e., respirometry or doubly labeled water (DLW)) to investigate behavioural variation and the energetic costs of hunting of completely free-ranging terrestrial carnivores. However, to date, most have focused on large carnivores specialized on large-bodied prey (e.g., cougars, Puma concolor (Linnaeus, 1771): Williams et al. 2014; leopards, Panthera pardus (Linnaeus, 1758): Wilmers et al. 2017; polar bears, Ursus maritimus Phipps, 1774: Pagano et al. 2018) and (or) species with particularly metabolically demanding hunting strategies (e.g., cheetahs, Acinonyx jubatus (Schreber, 1775), and African wild dogs, Lycaon pictus (Temminck, 1820): Gorman et al. 1998; Scantlebury et al. 2014; Hubel et al. 2016). Little is known about the interdependency of body size, activity, and metabolic requirements in free-ranging carnivores, particularly in the 10–15 kg size range.

Canada lynx (Lynx canadensis Kerr, 1792; hereinafter lynx) are mesocarnivores weighing ∼6 to 14 kg that prey on mammals and birds weighing <2 kg (i.e., much smaller than themselves; Fig. 1). Within the northern boreal forest, lynx specialize on snowshoe hares (Lepus americanus Erxleben, 1777), which are characterized by marked variation in abundance (i.e., 8- to 10-year population cycles; Krebs et al. 2018), but at peak densities are abundant on the landscape. Lynx populations vary according to the abundance of snowshoe hares, undergoing their own 8- to 10-year population cycle that lags 1 to 2 years behind snowshoe hare highs and lows (O’Donoghue et al. 1997, 1998a). Like most felids, lynx are generally solitary and considered ambush hunters spending much of their time at rest, ambushing prey from beds or stalking and chasing them for only short distances (Murray et al. 1995; O’Donoghue et al. 1998a). However, lynx are one of the few cold-climate members of the family Felidae, occupying the world’s coldest forested biome and one of the most seasonal environments. Although they are well adapted to their northern climate, including large, snowshoe-like feet providing low foot load for travelling in snow (Murray and Boutin 1991; Buskirk 2000) and thick, well-insulated winter pelts (Hammel 1955), their intermediate body size could impose cold-climate thermoregulatory demands, increasing the energetic cost of both resting and activity. Thus, lynx energetics are likely to be a compromise between a low-cost lifestyle enabled by ambush predation, a high-cost lifestyle imposed by a carnivorous diet and living in a cold, snowy climate, and a size-constrained lifestyle of being a large-bodied mesocarnivore reliant on small-bodied prey.

In this paper, we use animal-borne biologgers to document activity levels, movement, and heart rate alongside the first-ever estimates of daily energy expenditure (DEE) (using DLW: Speakman 1997) of free-ranging lynx in winter, during peak abundance of their primary prey (i.e., snowshoe hares). Our first objective was to assess the nature and degree of intraspecific and environmentally driven variation in energetics and behaviour of lynx. Based on interspecific allometries of physiological traits, we predicted that (i) larger bodied lynx would express higher activity levels and DEE (whole body expenditure), but lower heart rates (mass-specific expenditure) than smaller bodied lynx. Due to increasing thermoregulatory requirements in the cold, we predicted that (ii) resting heart rate would increase as air temperature (T_a) decreases and (iii) activity could either decrease to reduce movement costs in the cold or increase to generate heat and aid in thermoregulation. Finally, due to increasing costs of locomotion in the snow, we predicted that (iv) active heart rate and DEE would increase with increasing snow depth and (or) activity would decrease when snow was the deepest. Our second objective was to test the hypothesis of Carbone et al. (2007) that carnivores around the 14.5 kg threshold exhibit energy conservation strategies since they necessarily expend more energy than small carnivores, due to their body size, but are limited by the small size of their prey and (or) rates of processing or assimilation of food (i.e., intake is not unlimited). We test this hypothesis at (i) an interspecific level by comparing the energetic traits of lynx in relation to other carnivores above or below the 14.5 kg body mass threshold and (ii) at an intraspecific level by comparing how large- and small-bodied lynx in our study population situated around this threshold differ in energetic and behavioural traits.

All research was conducted along 50 km of the Alaska Highway, between Kluane Lake and Haines Junction, Yukon, Canada (61°N, 138°W), in a study area of approximately 300 km² (more details are available in section S.1.1 of the Supplementary material).¹ During our study period (November 2015 to April 2018), while densities of both snowshoe hares and lynx changed slightly from year to year, both species were relatively abundant (i.e., increasing and (or) at peak densities) in comparison with the low phases of their cycles (Krebs et al. 2019). Lynx are non-reproductive throughout the winter, but may start to breed into late March or early April (i.e., towards the very end of our collaring period). Male lynx tended to be solitary, whereas several of the female lynx that we collared were observed with offspring from the previous year and (or) other adult lynx. Snow was present on the ground between November and April in all years (mean depth of 25 cm; maximum depth of 69 cm). Mean monthly T_a ranged from –16.3 °C in December to –9.3 °C in March, and mean daily T_a ranged between –36.4 °C and 6.5 °C across all three study seasons.

By outfitting lynx with a combination of accelerometers, GPS collars, implantable heart rate loggers, and simultaneously obtaining DEE estimates via DLW, we reveal novel mass- and behaviour-related variation in lynx energetics and smaller-than-predicted influences of environmental variation. Lynx activity and movement were variable, but did not vary predictably with T_a, which varied by almost 40 °C, and was only slightly lower as snow got deeper. Heart rate increased slightly in the cold and with fresh snow when travelling, but again, with a lot of variation among individuals and across conditions. Consistent with interspecific expectations, the largest lynx were slightly more active and had lower heart rate (i.e., lower mass-specific metabolic rate). However, this was not reflected in DEE results, where, if anything, DEE (i.e., whole animal metabolic requirements) was lower for larger lynx. Collectively, these results indicate lynx, in general, and the largest individuals that we sampled (∼14 kg), in particular, minimize energy expenditure by maintaining very low activity levels, and while at rest, keeping costs low, which may be essential for this cold-climate, small-prey specialist that is too big to be an energetically efficient small carnivore.

Generally, lynx lived up to their reputation as an inactive, perhaps even “lazy” cat (Thompson 1977) spending, on average, only 21% of their day active. Lynx exhibited much lower rates of activity than similar-sized mesocarnivores (e.g., ocelot, Leopardus pardalis (Linnaeus, 1758); bobcat, Lynx rufus (Schreber, 1777); Iberian lynx, Lynx pardinus (Temminck, 1827); European badger, Meles meles (Linnaeus, 1758)) and sympatric boreal canids (e.g., red fox, Vulpes vulpes (Linnaeus, 1758); coyote, Canis latrans Say, 1823; wolf, Canis lupus Linnaeus, 1758). Their activity rates were more similar to the smallest bodied mustelid species, which escape to underground and subnivean spaces when inactive (Larroque et al. 2015), and the largest felid species, all of which are also ambush hunters, but specializing on large-bodied prey, and occupying warmer, highly productive environments. If lynx activity reflects the minimum amount of activity necessary to locate and capture enough prey to satisfy energy requirements (Curio 2012), then the low activity that we observed could have been accentuated by the high abundance of snowshoe hares, which were at peak densities in their population cycle during our study period (Krebs et al. 2019). Even so, based on interspecific comparisons, it seems that activity levels are dictated more strongly by hunting style than climate, making lynx more similar to other ambush predators than to sympatric, cold-climate cursorial predators.

There was substantial among-individual variation in lynx activity that was only weakly predicted by body size and not obviously responsive to environmental variation. In our study population, the largest bodied lynx had slightly higher activity levels and travelled greater distances in a day than smaller bodied lynx. However, plotting female and male lynx separately showed that this relationship is most evident in male lynx — larger males (∼13 kg) had higher activity than smaller males (∼9 kg) — but not in female lynx, which are smaller and show less size variation. Lynx were generally less active when snow was deepest, but did not vary activity according to T_a, despite a 40 °C range of T_a variation. These patterns mirror winter activity of snowshoe hares in the same location, which is constant across the same range of T_a (Menzies et al. 2020) and decreases when snow is deep (Peers et al. 2020). Surprisingly, while lynx and snowshoe hares are both assumed to be crepuscular and (or) nocturnal, we did not find a clear circadian pattern in the lynx activity or movement data (see section S.2.1 of the Supplementary material).¹ Previous studies have reported slightly higher activity in the late afternoons and early evenings, but not a nocturnal activity pattern (Kolbe and Squires 2007; Crowley et al. 2013). Overall, it appears that the daily routine of lynx activity (i.e., sleep–hunt–kill–eat–sleep, repeat) may not align with photoperiod or other environmental cues, but instead, correlate more strongly with spatial and temporal variation in prey behaviour and (or) be dictated by the energetic status of a given individual (e.g., hunting success, fasting tolerance, time since last meal; Podolski et al. 2013).

The mean heart rate of free-ranging lynx, across all individuals, days, and activity states, was 117 beats/min. Resting heart rate of free-ranging lynx averaged 94 beats/min and active heart rate averaged 144 beats/min (55% higher than resting levels). A major source of heart rate variation was interindividual variation in body mass, with smaller lynx characterized by faster heart rates than larger lynx. Across mammal species, from shrew to whale body sizes, large animals have larger hearts and slower heart rates than small animals (Williams et al. 2015). We show here that the same interspecific pattern in heart rate is present within the mammalian order Carnivora, declining from >200 beats/min in small mustelids to <100 beats/min in large felids, and within a single species. It is often the case that relationships become weaker or lower in magnitude as the taxonomic level considered is lower (i.e., moving from inter- to intra-order or from inter- to intra-species), but the observed negative relationship between body size and resting heart rate in lynx (decrease of ∼60 beats/min across 4 kg of size variation) was unexpectedly strong and steep compared with weak or non-existent intraspecific patterns found in the literature (Clark and Farrell 2011; Hezzell et al. 2013; Häggström et al. 2016). For example, Hezzell et al. (2013) document a difference of only 0.21 beats/min per kg of size variation, resulting in a 10.5 beats/min difference in heart rate for dog breeds ranging in size from 5 to 55 kg. So, while a negative relationship between heart rate and body size was expected, this relationship could be something more than a simple intraspecific manifestation of an interspecific pattern and, instead, an energy saving mechanism for larger, more active lynx. As well, felids tend to exhibit consistently lower heart rate than what is expected based on body size (i.e., to be “lion-hearted”), due to their low activity, low cost, low intensity ambush hunting styles (Williams et al. 2015), which our analysis shows to be true for lynx.

Beyond this size-related variation, there was evidence that lynx heart rate increased as T_a declined from cold to very cold, particularly when active, consistent with a thermoregulatory response to cold T_a. Results are consistent with a lynx lower critical temperature (T_LC) between –10 and –20 °C, most likely around –16 °C for resting lynx and –13 °C for active lynx. The only published estimate of lynx T_LC that we are aware of indicates that, for a single lynx weighing 12.6 kg, the thermoneutral extends to at least –10 °C, which was the lowest T_a that was measured (Casey et al. 1979). A previous study on bobcats, which are smaller bodied and not as cold adapted as lynx, estimated their T_LC to be –2.0 °C in winter (Mautz and Pekins 1989) and a study on red foxes in Alaska (USA) estimates their T_LC to be –13 °C (Irving et al. 1955). Finally, although lynx are efficient at moving through snow, we did observe a slight increase in active heart rate in response to greater accumulation of fresh snow, but more data, during periods with deeper snow and (or) greater variation in snow depth, could help investigate this further. Our biologging data shows stronger physiological responses than behavioural responses to T_a, a bit of both to increasing snow depth, neither in response to photoperiod, and both responding to variation in body size at the intra- and inter-specific levels.

Despite their low rates of activity and slow heart rates, lynx demonstrated slightly higher DEE than predicted for a carnivore of their body mass, with a DEE more comparable with similar-sized canids, which are typically highly active, cursorial species. The mean DEE of these 10 lynx, in winter, during peak snowshoe hare abundance, was ∼75 W; this rate of expenditure is equivalent to the energetic returns of killing 1.3 snowshoe hares per day (for calculations see section S.2.5 of the Supplementary material),¹ which is slightly higher but comparable with previously estimated daily kill or feeding rates of lynx at peak hare densities (1.2 snowshoe hares per day; O’Donoghue et al. 1998b; Studd et al. 2021). In general, cold-climate species tend to have higher upper boundaries of DEE than warm-climate species due to increased thermoregulatory requirements (Anderson and Jetz 2005). Thus, higher than expected lynx DEE may reflect its status as a cold-climate endotherm (White et al. 2007; White and Kearney 2013) and a vertebrate-specialist carnivore (Muñoz-Garcia and Williams 2005) studied during a period of abundant resources (Fletcher et al. 2012; Auer et al. 2016). Among-individual variation in DEE was most related to activity, but surprisingly, lynx that were less active tended to expend more energy. Although activity is more costly than resting (reflected in increased heart rate while active), the predominance of inactivity in lynx time budgets (∼80%) means that the cost of resting could be the primary determinant of DEE for lynx. And, as demonstrated by our biologging data, resting heart rate was higher for smaller lynx, which tended to be less active, and lower for the largest lynx, which tended to be more active. We predicted that larger lynx would have higher energy requirements, based on intraspecific (Konarzewski and Książek 2013) and interspecific (White and Kearney 2013) patterns in the literature, but if anything, our data suggests an opposite pattern (i.e., larger lynx having similar or lower energy requirements than smaller lynx). Although the negative relationship between heart rate and body size could reflect lower mass-specific energy requirements of larger animals, the magnitude of this pattern suggests that it could be more than that and could also reflect an energy conservation mechanism consistent with patterns seen in DEE data. There is evidence from other study systems that individuals with higher activity levels can compensate for greater activity by reducing expenditure during periods of rest, leading to reduced overall energy expenditure (evidence in birds; Welcker et al. 2015), which could also explain the negative relationship between DEE and activity (i.e., lower resting costs leads to lower expenditure despite greater activity). DEE also tended to increase with increasing snow depth, which is expected due to increased locomotory costs in deeper snow (Parker et al. 1984; Crête and Larivière 2003). We also observed in heart rate data (i.e., increased active heart rate with greater daily snow accumulation). We did not find significant effects of T_a on DEE in winter, but trends were consistent with the thermoregulatory costs suggested by heart rate responses (i.e., increased expenditure in the cold). We acknowledge that these DEE results must be interpreted with caution because they are based on limited sample size, only slightly significant or non-significant trends, and a novel fecal recovery method for DLW without an initial post-injection dilution estimate. Nevertheless, the results presented here represent the only estimates available for DEE of free-ranging lynx and do generally align with biologging data collected on more individuals and continuously over time.

Through a combination of emerging biologging technology and traditional measures of energy expenditure, we show the energy requirements of lynx appear to be driven by a complex interaction between the inactive, low-cost lifestyle of ambush hunters; the constraints of relying on small-bodied, mobile prey; and less prominently, the costs of living in an environment characterized by cold T_a and deep snow. The largest lynx that we studied (∼12–14 kg) are reaching the body size threshold determined by Carbone et al. (2007) at which feeding on smaller bodied prey becomes energetically infeasible. However, this infeasibility is defined primarily by the increased costs of activity experienced by larger bodied predators, without consideration of potential compensations while at rest. The predominance of inactivity in lynx time budgets (enabled by high resource abundance and short, highly successful prey pursuits), in addition to low costs while at rest (due to good insulation and low heart rates), might allow lynx to circumvent many of the costs that their large bodies and cold-climate existence otherwise impose on them. The question remains, whether the patterns that we demonstrate here — when hares are abundant and T_a is low — change after hares crash and lynx have to travel greater distances to find fewer hares, or when T_a becomes hot and large lynx have trouble dissipating heat. Further investigations into the seasonality and cyclicity, especially as it relates to the relative importance of intrinsic and extrinsic drivers of behaviour and energy expenditure, have potential to reveal greater insight into how predators in seasonal environments achieve positive energy balance at annual or multiannual scales, and whether size-driven, behaviourally driven, and environmentally driven energy economies remain the same over time.

A.K.M. and M.M.H. conceived the ideas and designed the study. S.B., D.L.M., and M.M.H. contributed to funding and logistics of field research. A.K.M., with the help of E.K.S., J.L.S., and RED livetrapped animals and collected the data. E.K.S. conducted behavioural calibrations of accelerometer data. A.K.M. conducted all laboratory work and other analyses. A.K.M. and M.M.H. led the writing of the manuscript. All authors contributed to interpretation of results, improving written drafts, and gave approval for submission.

Data will be made publicly available at the time of publication.

We are grateful to Agnes MacDonald and her family for long-term access to her trapline, as well as members of the Champagne and Aishihik First Nation for allowing us to conduct fieldwork as visitors within their homelands. We are thankful for the many lynx who carried our loggers and taught us so much. We thank numerous technicians and graduate students for help in the field, as well as Michelle Oakley, DVM, for her veterinary assistance and support on this project. We acknowledge local trappers (D. Drummond, L. Goodwin, L. Graham, S. Oakley, A. Preto) who helped inexperienced graduate students successfully livetrap lynx. Thank you go to Alina Evans and Asgeir Bjarnason for their help with implantable loggers, and to D. Powers, E. Ste Marie, F. Van Oordt, and K. Elliott for help with laboratory analyses. We thank M. Peers and Y. Majchrzak for the snow data and immense support in the field. Lodging and logistical support was provided by the Kluane Red Squirrel Project (KRSP). We also thank the Assiniboine Park Zoo for their support and help testing dataloggers in the initial phases of the project.