The mitochondrial physiology of torpor in ruby-throated hummingbirds, Archilochus colubris

Heterothermic mammals and birds avoid the cost of maintaining a high body temperature (T_b) during conditions of environmental stress (for example, cold exposure or low availability of food, water or O₂) by entering torpor. Torpor is a state of inactivity characterized by low metabolic rate (MR) and, usually, low T_b, which can be spontaneously reversed using endogenously produced heat (reviewed by Ruf and Geiser, 2015). In mammals, daily torpor is more common in small mammals such as Siberian hamsters (Phodopus sungorus, ∼30 g). In this species, MR typically drops by ∼70% and T_b decreases to as low as 15°C when housed at ambient temperatures of 15°C (Geiser, 2004). Seasonal hibernation is more common in larger mammals and is characterized by more profound and prolonged decreases in MR and T_b. For example, hibernating arctic ground squirrels (Urocitellus parryii, ∼800 g) decrease T_b to as low as −2.9°C, MR decreases by ∼95% and torpor bouts last up to 15 days (Buck and Barnes, 2000).

In mammalian heterotherms, suppression of mitochondrial metabolism mirrors the declines in whole-animal MR (Brown et al., 2007; Roberts and Chaffee, 1973). For example, in liver mitochondria isolated from torpid mice (Mus musculus) and Siberian hamsters (both daily heterotherms), phosphorylating respiration (OXPHOS, sometimes referred to as state 3) was suppressed by as much as 30% compared with that of normothermic, active animals, when measured at the same in vitro temperatures (Brown and Staples, 2011). The mitochondrial metabolic suppression observed in hibernators during the winter is even more dramatic; OXPHOS was suppressed by 70% in liver mitochondria isolated from torpid thirteen-lined ground squirrels (Ictidomys tridecemlineatus), again when measured at the same in vitro temperatures (Muleme et al., 2006). By contrast, changes in LEAK respiration (sometimes referred to as state 4) in torpor vary among mammalian endotherms depending on species. In liver mitochondria from Arctic ground squirrels, proton leak decreases when isolated from torpid animals compared with ‘active’ conspecifics, presumably reducing energy expenditure and decreasing heat production (Barger et al., 2003). State 4 respiration does not differ between torpor and arousal in liver mitochondria from another hibernator (thirteen-lined ground squirrels; Mathers and Staples, 2015) or a daily heterotherm (Djungarian hamster; Brown et al., 2007).

Many hummingbirds (avian family Trochilidae) are daily heterotherms that use torpor during cold nights (Hiebert, 1990; Lasiewski, 1963) particularly when fat stores are low, and in migration seasons torpor is used even when fat stores are large as a means to enhance fuel deposition (i.e. Eberts et al., 2021). Hummingbirds are an interesting model to study heterothermy as they have the highest mass-specific MR of any vertebrate, in part because they use hovering flight (Pearson, 1950; Suarez, 1992), yet they have the capacity to profoundly reduce T_b when needed. Hummingbirds are small (2–20 g) and have a high surface-area-to-volume ratio, so maintaining their typical daytime T_b of ∼39°C is energetically costly when ambient temperatures fall. Hummingbirds have some of the most profound decreases in MR during daily torpor, by as much as 95% (Wolf et al., 2020), on-par with mammalian hibernators. Hummingbirds in torpor share similar characteristics with mammalian heterotherms on the whole-animal level (decreased T_b, ventilation and heartrate; Withers, 1977), so we hypothesized that mitochondrial function is similarly suppressed.

Hummingbirds are nectivorous, and carbohydrates comprise most of their diet. While feeding during the day, hummingbirds' respiratory exchange ratio (RER) is near 1.0, which indicates carbohydrate catabolism. By contrast, even brief fasting (∼20 min after the last nectar meal) reduces RER to near 0.7, indicative of oxidation of stored fats (Groom et al., 2019; Krüger et al., 1982; Suarez et al., 1990). Similarly, at night when visual foraging is not possible, and during torpor, hummingbirds rely on stored fat as fuel (Hargrove, 2005; Powers, 1991). The regulation of this switch from carbohydrate to fat oxidation is mediated by the regulation of pyruvate dehydrogenase (PDH; reviewed in Zhang et al., 2014). PDH is inactivated by phosphorylation in the liver of Siberian hamsters when they enter daily torpor (Heldmaier et al., 1999), and is highly phosphorylated in the skeletal muscle of torpid mousebirds (family Coliidae; Green et al., 2022), another daily heterotherm. 3-Hydroxyacyl-CoA dehydrogenase (HOAD) is a rate-limiting enzyme in the β-oxidation pathway of lipid catabolism. HOAD activity increases in the liver of the marsupial Chilean mouse-opossum (Thylamys elegans) during arousal from torpor (Cortés et al., 2016) and in American goldfinches (Spinus tristis) during the winter (Marsh and Dawson, 1982), when both species require intense thermogenesis. Taken together, we hypothesized that the activities of these enzymes will be differentially regulated during torpor to support suppression of metabolism.

Ruby-throated hummingbirds are seasonal long-distant migrants that inhabit eastern North America and winter in South and Central America. They facultatively use torpor during cold nights and during migration (e.g. Eberts et al., 2021). We captured hummingbirds while they were summering in Ontario, Canada, induced torpor at night in a subset of individuals, and compared whole-animal and pectoralis mitochondrial metabolism as well as enzyme activities with those of normothermic birds. We used pectoralis as this tissue accounts for up to a third of hummingbirds' total body mass (Altshuler and Dudley, 2002), and muscle-specific mitochondrial MR in hovering hummingbirds is the highest among vertebrates (Suarez et al., 1991). The pectoralis is also a major thermogenic organ as birds do not have brown adipose tissue nor are they thought to use non-shivering thermogenesis (Marsh and Dawson, 1989), so we predicted suppression in mitochondrial function during torpor in this tissue might have a similarly large effect.

We hypothesized that, as in mammalian heterotherms, metabolic suppression of ruby-throated hummingbirds observed on the whole-animal level would be mirrored in mitochondria by suppressed OXPHOS. We evaluated oxidation of both carbohydrate and fatty-acid substrates and flux through both electron transport system (ETS) complexes I and II. We further hypothesized that the activities of relevant rate-determining metabolic enzymes would be altered in mitochondria to prioritize fatty acid metabolism during torpor. To our knowledge, these hypotheses have not been tested previously.

All protocols and procedures were approved by the University of Western Ontario Animal Care Committee (protocol number: 2022-028) and allowed under a Canadian Wildlife Service Scientific Collecting Permit (# SC-OR-2022-0256). We trapped wild male (N=5) and female (N=12; body mass 2.5–3.6 g) ruby-throated hummingbirds, Archilochus colubris (Linnaeus 1758), on the campus of the University of Western Ontario in London, ON, Canada (43.0°N, 81.3°W) from June to August 2022 and July to August 2023. The birds were captured using a trap-door-style hanging cage containing a hummingbird feeder using a method described by Eberts et al. (2021). Female hummingbirds were not captured until after 15 July in either year to ensure at least one breeding bout was complete. Once caught, the birds were transported to an animal facility on Western campus in cloth bird bags. The birds were housed individually in EuroCage enclosures (Corners Ltd, Kalamazoo, MI, USA), measuring 91.5 cm W×53.7 cm H×50.8 cm D. The birds were fed ad libitum on a 20% (w/v) solution of a Nektar-Plus (Guenter Enderle, Tarpon Springs, FL, USA) and sugar water (1:4 w/v) and were housed at a mean temperature of 24°C with 50% relative humidity. The birds experienced a semi-natural photoperiod that was adjusted weekly to match sunrise and sunset in London, ON, Canada.

We measured O₂ consumption and CO₂ production to quantify MR and determine when the birds were torpid. Torpor was induced by fasting the birds for 2 h before lights-out (Eberts et al., 2021), then placing them in the respirometry chamber within a temperature-controlled cabinet (Sanyo Biomedical) set to 10°C (actual chamber temperature 13°C). Measurements on normothermic birds were conducted in the daytime at least 1 h post-feeding in the respirometry system at a chamber temperature of 22.5°C.

Once steady state (torpid or normothermic for at least 1.5 h) was reached, the birds were quickly removed from the respirometry chamber and humanely euthanized by decapitation under isoflurane anesthesia, a common method of euthanasia in birds that is not known to affect mitochondrial activity (as in Coulson et al., 2024). Internal T_b was measured within 5 s of decapitation by inserting a thermocouple into the opened chest cavity of the bird. Then, the mass of the bird was recorded, and the pectoralis was dissected, weighed and divided into two portions. One portion was snap-frozen in liquid nitrogen for enzyme assays and the other was used for mitochondrial isolation. Total time from removal of the animal from the chamber to tissue freezing was approximately 10 min.

Pectoralis mitochondria were isolated from hummingbirds using differential centrifugation. A portion of tissue was weighed and placed in ice-cold biopsy preservation solution (BIOPS: 10 mmol l⁻¹ Ca-EGTA buffer, 0.1 μmol l⁻¹ free calcium, 20 mmol l⁻¹ imidazole, 20 mmol l⁻¹ taurine, 50 mmol l⁻¹ K-MES, 0.5 mmol l⁻¹ DTT, 6.56 mmol l⁻¹ MgCl₂, 5.77 mmol l⁻¹ ATP, 15 mmol l⁻¹ phosphocreatine, pH 7.1) during transit from the animal facility to the lab (approximately 30 min). To remove BIOPS, the tissue was rinsed with homogenization buffer [HB: 100 mmol l⁻¹ sucrose, 50 mmol l⁻¹ Tris, 10 mmol l⁻¹ EDTA, 100 mmol l⁻¹ KCl, 0.1% (w/v) BSA, pH 7.4 at 4°C] and minced. After mincing, the tissue was incubated with 0.2 mg ml⁻¹ proteinase (a non-specific protease) for 3 min, then rinsed with HB. We homogenized the tissue using a glass mortar and Teflon pestles at 100 rpm. The initial homogenization was done using a loose-fitting pestle (0.127 mm clearance), passing at least 10 times, until homogeneous. Then, a tight-fitting pestle (0.076 mm clearance) was used for two final passes. The homogenate was centrifuged at 4°C for 10 min at 1000 g. The pellet was discarded, while the supernatant was filtered through one layer of cheesecloth, and spun again at 4°C for 10 min at 8700 g. The supernatant was discarded, and the pellet re-suspended in HB and spun a final time at 4°C for 10 min at 8700 g. The supernatant was again discarded and the final pellet containing the mitochondrial fraction was re-suspended in 200 µl of HB. Mitochondrial protein concentration was determined using the Bradford assay (Bradford, 1976). Some mitochondrial protein was snap-frozen in liquid nitrogen and stored at −80°C for enzyme assays and the rest was used for high-resolution respirometry. Samples were stored on ice until respiration was determined (within 90 min of isolation).

The following substrates were added to measure OXPHOS capacity of fatty acids: 1 mmol l⁻¹ malate, 50 µmol l⁻¹ palmitoyl carnitine and 0.1 mmol l⁻¹ ADP. Once this rate was stable, complex V was inhibited with 1 μg ml⁻¹ oligomycin to estimate LEAK respiration (LEAK_Omy). Representative traces of these experiments are presented in Fig. S1. This protocol was first performed at 10°C with a subset of the isolated mitochondria, the machine re-calibrated, then performed again at 39°C. Preliminary experiments demonstrated that mitochondria did not lose activity over the 90 min total experimental time (data not shown).

The activities of the enzymes citrate synthase (CS) and HOAD were measured in duplicate using a SpectraMax M5 96-well plate reader (Molecular Devices) set to 39°C according to the protocol described by Price et al. (2010). Both enzymes were assayed in pectoralis tissue; CS was also measured in isolated pectoralis mitochondria. CS activity was measured over 3–5 min by the change in absorbance when DTNB is converted to TNB (absorbance at 412 nm); the reaction contained 50 mmol l⁻¹ Tris, 0.5 mmol l⁻¹ oxaloacetic acid, 0.15 mmol l⁻¹ DTNB, 0.15 mmol l⁻¹ acetyl-CoA and 3.4 µl of the homogenate diluted 1:10 for isolated mitochondria and 1:45 for the tissue. HOAD activity was measured over 3–5 min by the decrease in absorbance as NADH (absorbance at 340 nm) is converted to NAD⁺. The reaction contained 2 mmol l⁻¹ NADH, 10 mmol l⁻¹ EDTA, 10 mmol l⁻¹ acetoacetyl-CoA and 3.4 µl of the homogenate diluted 1:15.

PDH activity was measured using a coupled enzyme reaction (MAK183 PDH Activity Assay Kit, Sigma-Aldrich). The reaction (assay time, 30 min) results in NADH production proportional to PDH activity present in the homogenate. We attempted to measure maximal PDH activity by removing phosphate groups using alkaline phosphatase, but this measurement was incompatible with the kit and resulted in a complete loss of activity. Moreover, the very small amount of tissue available from each animal constrained our ability to troubleshoot this assay. As a result, we report only net PDH activity.

Whole-animal respirometry data were analyzed using ExpeData software (see Supplementary Materials and Methods for the annotated ExpeData macro). All statistics were modeled using GraphPad Prism software (version 5.00 for Windows, GraphPad Software Inc., San Diego, CA, USA). Data are expressed as means±s.e.m. To fit the criteria for the statistical models used in this study, all datasets were normal [D'Agostino-Pearson (omnibus K2) tests] and followed a Gaussian distribution (P<0.05). Because we had made directional predictions, one-tailed t-tests were used to compare means for V̇_O2, V̇_CO2, T_b, mitochondrial respiration rates and enzyme activities. Because RER is a ratio, those data were first transformed (logit) before a one-tailed t-test was performed. Comparison of temperature effects within individual samples was performed using two-way ANOVA with repeated measures, and the effect size of significant interaction effects between assay temperature and thermal state was determined. A value of P<0.05 was considered statistically significant.

We used whole-animal respirometry to confirm the metabolic state of the birds before tissue sampling. A bird with a high MR (V̇_O2≈0.004 ml min⁻¹) for at least 1.5–2 h was considered normothermic and a bird with low MR (V̇_O2≈0.0001 ml min⁻¹) for at least 2 h was considered torpid (Fig. 1).

Entrance into torpor reduced the MR (V̇_O2) by ∼95% (t=8.667, d.f.=9, P<0.0001; Fig. 2A). Likewise, V̇_CO2 was significantly reduced in torpid birds (by ∼96%, t=8.186, d.f.=9, P<0.0001). The RER did not differ significantly between normothermic (0.73±0.012) and torpid birds (0.77±0.032, t=2.079, d.f.=8, P=0.071, visualized within one individual in Fig. 1). T_b decreased significantly in torpor (13.1±2.0°C) compared with normothermia (37.1±2.1°C, t=23.34, d.f.=14, P<0.0001; Fig. 2B). The temperature coefficient of MR between normothermia and torpor (Q₁₀) was 2.84.

The rate of CI-supported OXPHOS respiration (OXPHOS_CI), stimulated using pyruvate, was 20.8% lower in torpor (114.0±16.6 nmol O₂ min⁻¹ mg⁻¹) than in normothermia (144.0±24.94 nmol O₂ min⁻¹ mg⁻¹, t=2.592, d.f.=11, P<0.05; Fig. 3A). There was no difference in estimated CII-supported OXPHOS (OXPHOS_CII; t=0.5174, d.f.=11, P=0.31. Fig. 3B) or in CI+CII-supported respiration (OXPHOS_CI+CII; t=1.368, d.f.=14, P=0.096; Fig. 3C). There was no difference in LEAK respiration rates supported by CI, CII or CI+CII (LEAK_CI, LEAK_CII, LEAK_CI+CII; t=0.2608, d.f.=11, P=3995; t=1.201, d.f.=11, P=0.1275; t=0.9734, d.f.=14, P=0.1734, respectively; Fig. 3D–F). There was no difference in respiration rate during OXPHOS or LEAK when respiration was fueled using palmitoyl-carnitine (t=0.8208, d.f.=11, P=0.2146 and t=0.7124, d.f.=9, P=0.4943, respectively; Fig. 4).

To evaluate temperature effects on mitochondrial respiration, we also measured respiration rates during OXPHOS and LEAK at 10°C, which approximates the T_b of torpid hummingbirds. We found that assay temperature significantly decreased mitochondrial respiration rates (P<0.0001) in both torpor and normothermia for all groups (Figs 3 and 4). However, there were interaction effects between metabolic state and assay temperature in both CI-supported (F_1,9=6.391, percentage total variation=1.842, P<0.05; Fig. 3A) and CII-supported OXPHOS (F_1,9=6.869, percentage total variation=7.390, P<0.05; Fig. 3B). Taken together, these results indicate that OXPHOS respiration fueled by pyruvate and succinate in mitochondria isolated from torpid animals is slightly, but significantly, less temperature sensitive than those from normothermic animals. Nonetheless, the average Q₁₀ for the pyruvate/succinate assays was 2.02 and did not differ between torpor and normothermia, apart from a significantly higher Q₁₀ of CI-supported OXPHOS in normothermia (2.7) compared with torpor (2.4; Table S1). The average Q₁₀ for the palmitoyl-carnitine assay was 1.8 and did not differ between torpor and normothermia (Table S1).

There was no difference between normothermia and torpor in CS activity in mitochondria isolated from pectoralis (normothermic: 1.846±0.13 µmol min⁻¹ mg⁻¹ mitochondrial protein, torpid: 1.8±0.22 µmol min⁻¹ mg⁻¹ mitochondrial protein, t=0.1391, d.f.=11, P=0.892; Fig. 5A) or in the pectoralis tissue (normothermic: 1371±139.5 µmol min⁻¹ g⁻¹ tissue, torpid: 1463±138.3 µmol min⁻¹ g⁻¹ tissue, t=0.4704, d.f.=8, P=0.651; Fig. 5B), indicating that mitochondrial quantity was unchanged between states. Neither HOAD activity (normothermic: 214.5±55.9 µmol min⁻¹ g⁻¹ tissue, torpid: 228.5±31.9 µmol min⁻¹ g⁻¹ tissue, t=0.2336, d.f.=10, P=0.820) nor PDH activity (normothermic: 8.953±1.302 mU mg⁻¹ mitochondrial protein, torpid: 7.461±1.048 mU mg⁻¹ mitochondrial protein, t=0.8923, d.f.=8, P= 0.199) in pectoralis tissue differed between states (Fig. 5C,D).

We also standardized mitochondrial respiration rates to the CS activity of isolated mitochondria. This analysis resulted in the same general trends described in Fig. 3: slight decreases in CI-supported respiration using pyruvate, but no differences in respiration with palmitoyl-carnitine (Fig. S2).

Combining data from both normothermic and torpid animals, we found that PDH activity positively correlated with CI+CII-supported OXPHOS at 39°C (R²=0.3847, P<0.05; Fig. 6A) and by CII-supported OXPHOS at 39°C (R²=0.6268, P<0.01; Fig. 6C). There was no correlation between CI-supported OXPHOS and PDH activity (R²=0.1595, P=0.143; Fig. 6B).

As has been reported in ruby-throated hummingbirds (Eberts et al., 2021) and many other hummingbird species (McKechnie et al., 2023; Ruf and Geiser, 2015) in both temperate and tropical zones (Pollock et al., 2022), we found that these birds entered torpor, where whole-animal MR declines sharply. Internal T_b also decreased close to ambient temperature (∼13°C), representing a decrease of nearly 25°C. In six species of torpid Andean hummingbirds at natural ambient temperatures (2.4–5.9°C) T_b dropped between 26 and 30°C (Wolf et al., 2020); that study also reported the lowest recorded torpid T_b of 3.26°C in black metaltail (Metallura phoebe).

Our study design allowed us to assess metabolism at levels from the whole animal down to enzymes in the same individuals. Contrary to our hypothesis, we found that changes in mitochondrial metabolism did not mirror those at the whole-animal level: OXPHOS was not lower in torpid hummingbirds compared with normothermic birds when measured in isolated mitochondria at 39°C. As shown in our data and other studies (Hiebert, 1990; Krüger et al., 1982), fat is the primary fuel source hummingbirds use during fasting and torpor, but we found no change in the capacity for maximal mitochondrial oxidation of fatty acid substrates during torpor. There was a ∼20% decrease in CI-supported respiration fueled with pyruvate, but the physiological relevance of this effect is unclear, as the RER showed that carbohydrates were not oxidized in vivo. These results were surprising considering that suppression of mitochondrial metabolism in torpor is so well categorized in mammalian heterotherms (reviewed by Staples, 2014). In mammalian daily heterotherms, whole-animal MR decreases by ∼70% and OXPHOS by ∼30% (as in Brown and Staples, 2011). However, despite the suppression of whole-animal MR by 95%, we did not observe proportional decreases in OXPHOS. We also did not see changes in LEAK respiration between normothermic and torpid hummingbirds, indicating that the coupling between proton pumping and ATP synthesis remained steady.

We chose pectoralis as this tissue is a major thermogenic organ in avian species that contributes disproportionately to whole-animal MR (Suarez et al., 1991), so we predicted that any changes in mitochondrial respiration would be greatest there. In mammalian heterotherms, OXPHOS in mitochondria isolated from skeletal muscle also decreases significantly in torpor (Barger et al., 2003; Brown et al., 2012; James et al., 2013), though not to the extent in tissues such as the liver that are more metabolically active in resting animals (reviewed in Staples and Brown, 2008). It would be interesting to repeat this study using liver or heart, but very small tissue sizes may require pooling of samples.

As the T_b of torpid and euthermic hummingbirds differed (∼37°C versus ∼13°C), we assessed respiration at two assay temperatures approximating each physiological steady-state temperature. We observed lower mitochondrial respiration in torpid hummingbird mitochondria when measured at 10°C (an assay temperature approximating torpid T_b), compared with that at 39°C (approximating normothermic T_b) in all respiratory states tested. This finding is probably due to passive thermal (i.e. Arrhenius) effects. The Q₁₀ for respiration rates was ∼2 in all cases, similar to the Q₁₀ of whole-animal MR (2.84). As torpid birds have a T_b near 10°C, we assume that ATP supply rate is significantly reduced in this state, although this does not indicate an active suppression in mitochondrial function as seen in mammalian heterotherms, where suppression is observed independently from assay temperature (i.e. Brown and Staples, 2011). A recent study by García-Díaz et al. (2023) tested for effects of mild hypothermia induced by low night-time temperatures on blood cell mitochondrial respiration in great tits (Parus major, not known to use torpor). They also reported a Q₁₀ effect (in vitro temperature-dependent decreases in OXPHOS), similar to what we report here.

CS activity, a common biomarker for mitochondrial quantity in skeletal muscle (Larsen et al., 2012), did not differ between torpid and normothermic hummingbirds in isolated mitochondria and in whole pectoralis. Not surprisingly, this indicates that the apparent abundance of mitochondria does not change over the short transition from normothermia to torpor. We standardized OXPHOS and LEAK measurements to CS activity (as opposed to per milligram of mitochondrial protein, as in Mogensen et al., 2006) and found the same patterns as when respiration was quantified per milligram of mitochondrial protein.

It is likely that changes in post-translational modifications to PDH largely accounted for the suppression in carbohydrate-based respiration we observed in torpor (as it does in torpid mousebirds; Green et al., 2022). Although we did not find statistically significant differences in PDH activity between states, OXPHOS rates though CI and CII did correlate positively with PDH activity. Unfortunately, we were unable to measure PDH phosphorylation state. Both our normothermic and torpid animals were in a fasted state when they were sampled, and RER values suggest exclusive lipid oxidation, so it is likely that some changes in PDH regulation occur simply as a result of fasting. However, further changes may also occur as hummingbirds enter torpor. It would be interesting to separate the effects of fasting from torpor, but hummingbirds offered nectar would not enter torpor in our experiment.

Our measured HOAD activity is compatible with that reported in the pectoralis of rufus hummingbirds (Selasphorus rufus; Suarez et al., 2009) and four other hummingbird species (Fernández et al., 2011). HOAD activity was unchanged between the steady states of torpor and normothermia, though it would be interesting to determine whether this enzyme is upregulated during the arousal period where rewarming occurs rapidly (within 15 min), as it is in mouse-opossums (Cortés et al., 2016) and shivering goldfinches (Marsh and Dawson, 1982).

Heterothermy is thought to be a conserved ancestral trait in mammals (Geiser, 1998), but a later adaptation in birds. Whereas a wide variety of small marsupial, placental and monotreme mammals use deep torpor extensively, hummingbirds, swifts and nightjars are the only known types of birds that use this thermal strategy (reviewed by McKechnie et al., 2023). In many of the mammalian heterotherms studied so far, active, temperature-independent suppression of mitochondrial metabolism is a common characteristic of the torpor phenotype. By contrast, the data presented herein seem to indicate that the mechanisms used by hummingbirds to actively suppress MR during torpor differ from those of mammalian heterotherms, a finding that, to our knowledge, has not yet been reported. Alternatively, another cellular mechanism hummingbirds may use to decrease MR in torpor is suppression of the pathways of ATP demand. In future work, we will measure the activity of major consumers of ATP, such as sarcoendoplasmic reticulum calcium ATPase (SERCA) and Na⁺/K⁺-ATPase. In fact, the activity of Na⁺/K⁺-ATPase decreases by 60% in the skeletal muscle of the golden-mantled ground squirrel (Callospermophilus lateralis) in hibernation (MacDonald and Storey, 1999).

It is possible that the extreme decrease in whole-animal MR and T_b of hummingbirds is achieved by passive cooling alone. Hummingbirds are small-bodied birds and typically live at ambient temperatures well below their thermoneutral zone (TNZ). Shankar et al. (2019) estimated a TNZ of 32–35°C for the 3–4 g broad-billed hummingbird (Cynanthus latirostris), so we assume a similar TNZ for the 2.5–4 g ruby-throated hummingbirds. When the birds used in this study arrive from spring migration to their summering habitat in London, ON, Canada, in May, they might experience average night-time temperatures as low as 9°C (https://climate.weather.gc.ca/climate_normals/index_e.html, accessed 3 May 2024). As hummingbirds regularly experience ambient temperatures as much as 30°C below their normothermic T_b, we predict that passive cooling could occur quickly and provide a mechanism to rapidly suppress MR through passive thermal effects. The Q₁₀ for whole-animal metabolism was similar to those reported for other small avian daily heterotherms (Geiser, 1988). Moreover, the Q₁₀ for mitochondrial respiration rates fell within the range of 2–3 expected for passive thermal effects on reaction rates. So, simply cooling these organelles during entrance into torpor likely allows hummingbirds to benefit from the energy savings associated with decreased ATP production. In hummingbirds, metabolic suppression is probably achieved by decreasing thermoregulatory setpoint (reviewed in Staples, 2014) to a low, torpid T_b. With the large energy savings achieved by this route, further active suppression of mitochondrial metabolism is likely to be of little benefit.

The authors thank Sharla Thompson for her help with setting up animal housing equipment and overseeing daily checks, Dr Morag Dick, Dr Catherine Ivy and Kevin Young for their guidance with whole-animal respirometry, and Soren Coulson for his advice on enzyme assays.