Torpor use in the wild by one of the world’s largest bats

Abstract

Torpor is widespread among bats presumably because most species are small, and torpor greatly reduces their high mass-specific resting energy expenditure, especially in the cold. Torpor has not been recorded in any bat species larger than 50 g, yet in theory could be beneficial even in the world’s largest bats (flying-foxes; Pteropus spp.) that are exposed to adverse environmental conditions causing energy bottlenecks. We used temperature telemetry to measure body temperature in wild-living adult male grey-headed flying-foxes (P. poliocephalus; 799 g) during winter in southern Australia. We found that all individuals used torpor while day-roosting, with minimum body temperature reaching 27°C. Torpor was recorded following a period of cool, wet and windy weather, and on a day with the coldest maximum air temperature, suggesting it is an adaptation to reduce energy expenditure during periods of increased thermoregulatory costs and depleted body energy stores. A capacity for torpor among flying-foxes has implications for understanding their distribution, behavioural ecology and life history. Furthermore, our discovery increases the body mass of bats known to use torpor by more than tenfold and extends the documented use of this energy-saving strategy under wild conditions to all bat superfamilies, with implications for the evolutionary maintenance of torpor among bats and other mammals.

1. Introduction

Torpor is a physiological state of energy conservation used by a wide range of mammals and birds [1]. During torpor, body temperature and metabolic rate are reduced below levels maintained in a normothermic state [2]. Torpor differs from hypothermia because it is controlled and reversible by endogenous heat production [3]. Torpor can substantially decrease an individual’s rate of resting energy expenditure and is used in response to a decrease in food availability [4], such as during winter [5], adverse weather events [6] or after wildfires [7]. Torpor is also used when food is available but relatively costly to obtain [8], including because of interspecific competition [9] or risk of predation [10,11]. The relative energy savings gained from torpor are greater for smaller species because of their higher mass-specific rates of heat loss and hence metabolic heat production to maintain normothermia [2], and accordingly, torpor appears more common in smaller mammals and birds [1].

Most bats are relatively small (figure 1), with a median body mass of just 12 g [14]. Their small size has been linked to constraints on evolutionary selection imposed by flight [15] and echolocation [16], with ancestral bats hypothesized to also weigh just 12 g [14]. Torpor use presumably confers a fitness advantage to small bats because it reduces their energy budgets to levels that are less extreme relative to their body mass [17], and accordingly, torpor has not been recorded in any bat species weighing more than 50 g [18]. The high mass-specific rate of resting energy expenditure of small bats [19] thus provides a compelling explanation for the widespread use of torpor among this group. Alternatively, because food availability can be inconsistent depending on seasonal or sporadic environmental conditions, periods of severely reduced food energy intake could still favour the evolutionary maintenance of torpor even in larger bat species, and especially in those species that inhabit colder, more seasonal climates.

Figure 1.

The distribution of average body mass for 656 species of the order Chiroptera (data taken from [12]). The arrow indicates the mean body mass of the grey-headed flying-fox (Pteropus poliocephalus; 799 g) (taken from [13]). Image: J.A. Welbergen.

Whereas most bats are relatively small, some flying-foxes (Pteropus spp.) have average species body masses of up to 1200 g [12], with some individuals weighing up to 1600 g [20]. Among all fruit bats (family Pteropodidae), torpor has been recorded in only a few relatively small (<50 g) species and only under captive conditions [21,22]. Moreover, the few studies using telemetry to record thermoregulatory patterns of wild-living fruit bats (Nyctimene robinsoni 60 g [23], Rousettus aegyptiacus 135 g [24]) did not find torpor use. Torpor has not been reported in any flying-fox (i.e. Pteropus) species [18,25–28]. Nevertheless, the large size of flying-foxes compared with most bats (figure 1) provides an opportunity to test the hypothesis that environmental and ecological factors are predictors of torpor use among bats, irrespective of body mass. Whereas many flying-fox species experience tropical and oceanic climates, some are found at higher latitudes where they must cope with strong seasonal and day-to-day variation in environmental conditions. Moreover, because they roost in the tree canopy on exposed branches, flying-foxes are subjected to the full extent of daily fluctuations in environmental thermal conditions. In addition, flying-foxes rely on an ephemeral diet of nectar, pollen and fruit, and can experience increased starvation risk when these food resources are limited, especially during periods of cold and severe weather [29–31]. Like colder, more seasonal conditions, variable food availability is a strong predictor of torpor use [4], including in nectivorous birds [32] and mammals [1]. Thus, flying-foxes face similar conditions that favour torpor in other species, but until now this energy-saving mechanism has not been demonstrated in this group.

Here, we used temperature-sensitive telemetry to continuously record the core body temperature of adult, male grey-headed flying-foxes (Pteropus poliocephalus) while day-roosting in the wild during winter in southern Australia. This species occurs along the southeastern coast of Australia from subtropical to temperate regions [13] where they occupy higher latitudes than any other Pteropus species globally [33], and our study was conducted at one of the most southerly roost sites of this species [34]. During summer, grey-headed flying-foxes can experience extremely hot conditions that can cause mortality [35,36]; whereas, during winter, anecdotal evidence indicates that starvation and cold exposure can also cause mortality [29–31]. Adult male grey-headed flying-foxes reach an annual minimum in their body condition during late autumn and early winter [37]. We predicted that male grey-headed flying-foxes can use torpor while day-roosting, particularly in response to energetic stress imposed by cold weather during winter. The results of our study are important for understanding the physiological mechanisms that enhance survival in this ecologically valuable mammal group and, more broadly, the evolution and expression of torpor among bats and all mammals.

2. Methods

We studied the grey-headed flying-fox (Pteropus poliocephalus 799 g) [13] at a roost site in the city of Adelaide, South Australia (−34.916° S, 138.607° E). Bats were caught using mist nets set at canopy height with a rope pulley system prior to dawn in June 2021 and transmitters were implanted in seven adult males (body mass: 791 ± 80 g). Captured bats were placed in cotton bags and moved to the nearby Animal Health Department at Adelaide Zoo. During a short surgery under isoflurane-induced anaesthesia, we implanted a sterilized temperature-sensitive radio transmitter (model PD-2THX, 3.9 g; Holohil, Ontario, Canada) into the peritoneal cavity of each bat. Each transmitter was pre-calibrated in a water bath against a National Institute of Standards and Technology (NIST) traceable digital thermometer (Control Company, Texas, USA). During surgery, we injected subcutaneously carprofen (3 mg kg⁻¹; Rimadyl, Zoetis Inc., Kalamazoo, USA), a non-steroidal anti-inflammatory drug, and Hartmann’s fluid (20 ml). We held bats following surgery in individual cages at room temperature with food (fruit) and water overnight and the next day after an assessment of wound healing released them back into the Adelaide roost.

To continuously record transmitter pulse interval and hence core body temperature of roosting bats, we used two autonomous data-logging receiver stations (model R4500S, each connected via a switch box to two directional antennas; Advanced Telemetry Systems Australia, Gold Coast, Australia) located around the periphery of the roost. Data were recorded by the receiver stations every 10 min when bats were within the range of reception (approx. 200 m) of the implanted transmitters, which included most of the area used for roosting by bats. We recorded very few data for one individual, presumably because it often roosted outside of the range of the data loggers, and therefore, we removed this individual from the analysed dataset. For the remaining six individuals, we removed data for the first 2 days after release to account for any possible effects of captivity and the last 10 days of recorded data to remove any possible effect of reduced transmitter battery voltage on calibration relationships (see electronic supplementary material, figure S1 for the entire dataset). Our final analysed dataset included 13 023 measurements of core body temperature recorded from six individuals (577 to 3866 values per individual) over 44 days between 20 June and 3 August 2021 (austral winter). We sourced data on air temperature (°C), wind speed (km h⁻¹) and precipitation (mm accumulated over 1 min) from a weather station (#023000) maintained by the Australian Bureau of Meteorology located 2.9 km west of the roost.

We analysed data using the software programs R [38] interfaced with RStudio [39]. Results are presented as the mean ± 1 s.d. of within-individual mean or, for times relative to sunrise and sunset, median, values. To delineate torpor, we used a conservative threshold of <5°C below an individual’s normothermic body temperature [40,41], which we estimated from the mode of their frequency distribution of day-roosting body temperature. Maximum rates of cooling and rewarming were evaluated over a 10-min period. Torpor bout duration was measured as the time body temperature remained below the individual torpor threshold. To visualize the temporal pattern and variation in body temperature we plotted the data as a frequency histogram binned into each hour of the day. To investigate how environmental conditions influenced the regulation of body temperature, we fitted a linear mixed effects model to the observed variation in body temperature. This global model included as fixed effects: air temperature, wind speed, precipitation and two-way interactions among these three weather variables. We included as a factor type variable hour of the day to account for the possible influence of an endogenous circadian rhythm. Finally, we included individual identity as a random effect on the intercept to account for variation in average body temperature among individuals and the within-individual repeated measures structure of the data. Models were fitted using the function ‘lmer’ from the package ‘lme4’ [42]. A list of models including all combinations of fixed effects in the global model were fitted using the maximum likelihood (ML) method and ranked by Akaike information criterion (AICc) using the function ‘dredge’ from the package ‘MuMIn’ [43]. The most parsimonious model was selected as that with the fewest variables from models with an AICc value more than 2 units less than other models (i.e. a variable was retained only if it decreased the AICc by >2 units). This final model was fitted using the restricted maximum likelihood (REML) method with significance tests for effects implemented using the package ‘lmerTest’ [44]. We fitted a segmented regression to the relationship between wind speed and body temperature using the package ‘segmented’ [45]. Predicted effects and 95% confidence intervals (in the form of estimate ± multiple of the standard error) were predicted and visualized using the package ‘visreg’ [46]. Model residuals were checked for approximate normality by inspection of histogram, partial effects and quantile–quantile plots.

3. Results

During the winter study period, air temperature ranged daily between 8.4 ± 2.0°C (min.: 4.2°C) and 15.2 ± 1.9°C (max.: 19.1°C). The lowest daytime maximum temperature was 9.1°C, occurring on 22 July. Wind speed, averaged over the day, reached a maximum of 23 km h⁻¹, and rain occurred during some periods (figure 2).

Figure 2.

(a) Body temperature (°C) recorded in a wild-living adult male grey-headed flying-fox (Pteropus poliocephalus) during winter at the roost site, (b) air temperature (°C), (c) wind speed (km h⁻¹) and (d) occurrence of rain (yes = 1 or no = 0) recorded at a nearby weather station. The horizontal black line shown in panel (a) highlights a period in mid-winter when all six implanted bats exhibited a greater than usual decrease in body temperature during resting. At the beginning of this period, bats experienced several days of relatively high wind speed (c), frequent rain (d) and a decrease in air temperature (b). On 22 July (within vertical grey lines), bats exhibited a particularly low body temperature (min.: 27°C), indicative of torpor, while experiencing the coldest daytime maximum air temperature recorded (9.1°C), moderate wind speed and rain during day-roosting.

Body temperature was recorded only while implanted bats were at the roost site (electronic supplementary material, figure S1). The timing of arrival and departure of bats at the roost site, as indicated by the first and last measurements of body temperature for five out of six individuals for which we often had continuous data during day-roosting, was 50 ± 21 min before sunrise and 36 ± 8 min after sunset, respectively. Body temperature over the day-roosting period was 37.0 ± 0.3°C and varied between a minimum of 35.1 ± 0.2°C and maximum of 39.0 ± 0.5°C (figures 2 and 3). The minimum recorded body temperature was 27.0°C (30.2 ± 2.2°C among individuals), which was 11.0°C (7.5 ± 2.3°C among individuals) less than the mode of day-roosting body temperature data recorded for this individual of 38.0°C (37.7 ± 0.4°C among individuals; electronic supplementary material, figure S2). The minimum recorded body temperature exceeded >5°C below normothermic body temperature (i.e. the mode), a threshold we chose to delineate torpor, for all individuals. For the two individuals with sufficiently continuous data, the torpor bout observed on the morning of 22 July lasted 5.0 and 3.0 h, respectively. Entry into torpor occurred at a maximum rate of cooling of −0.1 ± 0.04°C min⁻¹ and arousal occurred at a maximum rate of rewarming of 0.2 ± 0.07°C min⁻¹.

Figure 3.

(a) and (b) Body temperature (°C) of two wild-living adult male grey-headed flying-foxes (Pteropus poliocephalus) at the roost site and (c) air temperature (°C) over a period of 5 days in winter, including when the lowest body temperature was observed in all individuals.

A decrease in body temperature was commonly observed in the morning, with minimum daily body temperature occurring between 08.00 and 11.00. Body temperature was also more variable during the morning compared with after 12.00 (figure 4). Another decrease in body temperature typically occurred in the late afternoon between 17.00 and 18.00, with minimum body temperature recorded after 12.00 as low as 32.5°C (33.1 ± 0.6°C among individuals), prior to bats leaving the roost (figures 3 and 4). Relatively high body temperature values occurred at the time of arrival into the roost site, which spanned between 05.00 and 06.00, and over the middle of the day around 13.00 (figure 4).

Figure 4.

Frequency distributions (black shapes) and inserted boxplots showing the median (black dot) and 25th and 75th percentiles (white box) of body temperature (°C) recorded within each hour of the day for six adult male grey-headed flying-foxes (Pteropus poliocephalus) at the roost site in winter.

A linear mixed effects (LME) model fitted to explain variation in body temperature included the fixed effects of hour of the day, a negative effect of rain occurrence and an interaction between a positive effect of air temperature fitted as a quadratic relation and a logical (yes/no) wind variable designating whether wind speed was less than or greater than 13.3 km h⁻¹ (table 1; figure 5). The fitted curvilinear effect of air temperature indicated a stronger effect on body temperature at the lower range of air temperatures experienced (below approx. 12°C), which was even stronger at times of relatively high (i.e. >13.3 km h⁻¹) wind speed (figure 5). A segmented regression analysis of the linear relationship between body temperature and wind speed indicated a significant difference in slope at 13.3 km h⁻¹ (Davies’ test: p < 0.0001), and a transformation to a logical variable simplified fitting and interpretation of the interaction between wind speed and air temperature effects on body temperature.

Table 1.

Analysis of deviance table (type 3 Wald chi-square tests) for a linear mixed effects model fitted to explain variation in body temperature (13 020 observations from six individuals) of adult male grey-headed flying-foxes (Pteropus poliocephalus) at the roost site. A random effect on the intercept fitted for individual identity was associated with a variance of 0.02 (residual variance: 1.30). Marginal/conditional R ²: 0.35/0.36.

fixed effects	chi-sq.	d.f.	p‐value
intercept	1417	1	<0.0001
wind	264	1	<0.0001
air temp2	43	1	<0.0001
air temp	158	1	<0.0001
rain	23	1	<0.0001
hour of day	3106	19	<0.0001
wind:air temp2	122	1	<0.0001
wind:air temp	176	1	<0.0001

Figure 5.

Predicted effect of air temperature (lines and shaded 95% confidence intervals) from a linear mixed effects model and measured body temperature values (symbols) under conditions of relatively low or high wind speeds (< or > 13.3 km h⁻¹) for adult male grey-headed flying-fox (Pteropus poliocephalus) at the roost site in winter. The model also included fixed effects of hour of the day and rain (yes/no), and a random effect on the intercept of individual identity. See table 1 for model results.

4. Discussion

Our study is the first to measure body temperature in a wild-living flying-fox (Pteropus spp.) and the first to record torpor use by a fruit bat (Pteropodidae) under natural conditions. Our results increase the body mass of bats known to use torpor more than tenfold to now include one of the world’s largest bats and extend the documented use in the wild of this energy-saving strategy to all bat superfamilies. Evidently, torpor among bats is not merely a function of their mostly small body size. In our study, torpor use by the grey-headed flying-fox coincided with low air temperature, rain and high wind speeds, and at a time of the year when prior studies have shown male body condition is recovering from a seasonal low [37], suggesting that torpor was used to reduce energy expenditure during periods of increased thermoregulatory costs and depleted body energy stores. Our discovery has implications for understanding the distribution, behavioural ecology and life history of flying-foxes, as well as the evolutionary maintenance of torpor among bats and other mammals.

In our study, minimum body temperature exceeded a threshold for delineating torpor of >5°C below normal resting values [41] and the dynamics of body temperature change also were indicative of torpor bouts rather than uncontrolled hypothermia. The timing of torpor in the morning matched the daily pattern of cooling on other days, and corresponds with a strong endogenous rhythm for torpor in the morning commonly found among smaller insectivorous bats [47]. Cooling resembled that of a mammal entering torpor, when normothermic thermoregulation is temporarily de-activated because of a lowered thermoregulatory set-point [48], with rapid initial cooling and a maximum rate that exceeded that predicted (0.04°C min⁻¹) for a similar-sized mammal entering torpor [41]. In contrast, cooling in a hypothermic mammal begins gradually and increases as a colder body temperature suppresses metabolic heat production [3]. Rewarming to normothermia occurred despite an increase in air temperature of only 3°C on this day, indicating that it was achieved by endogenous thermogenesis. Maximum rates of rewarming were close to the rate predicted (0.21°C min⁻¹) for placental mammals arousing from torpor [49]. Together, acknowledging that metabolic rate is required to distinguish torpor from hypothermia unambiguously [3], these aspects of body temperature provide convincing evidence for a controlled state of torpor.

Torpor was used by male grey-headed flying-foxes only during a period of adverse weather conditions, rather than being used routinely during roosting as found in many small bats [18]. In all measured individuals, torpor was preceded by days when body temperature was decreased in the morning below normal levels observed on most days in winter. A similar pattern was found in at least one other heterothermic mammal (Sugar glider, Petaurus breviceps, 125 g) [50,51]. The larger than normal decrease in body temperature during resting on preceding days is typical of endotherms deprived of food and in a state of energy stress [52]. The flying-foxes in our study could have been energetically stressed in the period leading up to torpor because of the cold air temperature, consistent rain and high wind speed. These conditions increase heat loss and thermoregulatory energy costs during resting and foraging [27], and reduce eucalypt nectar volume and sugar concentration [53], possibly leading to lesser rates of food energy intake. Torpor at the observed minimum body temperature of 27°C and air temperature of 6°C could reduce the resting metabolic rate of a torpid but thermoregulating (as opposed to a thermoconforming) bat to approximately 50% of normothermic levels if we assume a minimum thermal conductance measured in normothermic bats [27]. However, energy savings were possibly greater because of increased normothermic costs under natural conditions during exposure to wind and rain. Hence, the maximum 5.5 h torpor duration, representing about half the mid-winter daytime period at this location, would provide a meaningful reduction to rest-phase energy costs. The energy required for rewarming, although comprising a large proportion of the energy used over a torpor bout [54], does not negate the energy savings gained from torpor because the energy cost of arousal is similar to that otherwise required to maintain normothermia over the time taken for arousal [55]. Interestingly, we found that bats began nocturnal activity (flying out of range of the receiver/dataloggers) before rewarming from a common late afternoon decrease in body temperature, presumably avoiding the cost of rewarming by using heat generated by activity. Flight at a relatively low body temperature as bats rewarm from torpor has been reported in other smaller species [56]. By reducing body temperature during resting, including entering torpor bouts, an individual reduces their energy cost for resting and hence lowers their risk of starvation and mortality when energetically stressed.

Our observations expand our knowledge of torpor under wild conditions to even the largest bats and to all the bat superfamilies. The evolution of bat species within the Pteropodidae is associated with large and rapid changes in body size [14] and a high rate of diversification [57]. Given that torpor has now been recorded across the full range of body masses and in phylogenetically distant genera within the Pteropodidae (this study, [22,58,59]), it is parsimonious to assume that torpor is a shared ancestral trait among all pteropodids, and indeed all bats, given the early divergence of Pteropididae from other bats [60], with its expression among species likely dependent on intrinsic (e.g. body mass, roosting behaviour) and environmental factors (e.g. seasonality, extreme events) that influence the risk of energy stress and starvation mortality. For example, many flying-fox species occur on tropical islands with a warm and oceanic climate but a capacity for energy-saving torpor could still be advantageous during periods when nectar or fruit are limited, including during and in the aftermath of severe tropical storms. Frugal energetic strategies appear to facilitate the persistence of island species [61]. As the most important physiological mechanism for reducing energy expenditure, torpor could be a critical trait explaining the remarkable ability of flying-foxes to persist within relatively small island habitats [62], and for this reason, torpor might also be linked to their high rate of diversification [63]. It remains to be formally tested, however, whether other flying-fox species occupying tropical island habitats can use torpor. Finally, given that bats diverged from other mammals around 58 million years ago [64,65], evidence of torpor expression in the wild among all major bat groups provides additional support for torpor being a plesiomorphic trait that evolved in concert with the development of endothermy in early mammals [66,67].

Ethics

All procedures were approved by the Animal care and Ethics Committee of Western SydneyUniversity (A12217) and a permit granted by the Government of South Australia (M26735-3).

Data accessibility

Data are available at Dryad [68].

Supplementary material is available online [69].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors’ contributions

C.T.: conceptualization, formal analysis, funding acquisition, investigation, visualization, writing—original draft, writing—review and editing; M.W.: investigation, writing—review and editing; W.B.: investigation, writing—review and editing; J.M.M.: investigation, writing—review and editing; A.M.: investigation, writing—review and editing; J.M.: funding acquisition, investigation, writing—review and editing; J.A.W.: funding acquisition, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.