Introduction

Torpor is characterised by pronounced reductions in body temperature (T b), metabolic rate (MR) and water loss (Hock 1951). Torpor is a key strategy employed by many mammals to conserve energy and water and to overcome adverse environmental conditions. Low ambient temperature (T a) or a decrease in food availability are often prerequisites for use of torpor (Wang 1989). However, individuals of some species regularly enter torpor even when food is available and weather conditions are mild (Turbill et al. 2003). In these individuals, torpor appears to be part of a normal daily routine that is used to manage energy and/or water requirements.

Torpor is predominantly employed by small mammals, which, because of their large surface area to volume ratio, require large amounts of energy relative to their size to maintain an elevated warm T b (normothermia). Microchiropterans (microbats) are among the smallest mammals (most of them weigh <25 g) and because many species feed on insects, their energy supply and requirements are strongly affected by T a (Fenton 1983). The use of torpor to cope with low food availability is widespread among microbat species belonging to several families and inhabiting a range of climates (Henshaw 1970; McManus and Nellis 1971; Genoud et al. 1990; Arlettaz et al. 2000; Geiser and Brigham 2000; Turbill and Geiser 2006; Willis et al. 2006; Wojciechowski et al. 2007).

Most work on thermal biology has been conducted on temperate-zone bats, which generally employ short bouts of torpor lasting less than a day during the warmer months and prolonged multi-day torpor during winter hibernation (Speakman and Thomas 2003; Geiser 2004). In contrast, knowledge about the thermal biology of free-ranging bats in subtropical or tropical habitats, with vastly different environmental conditions and energetic constraints, is scarce. However, even in tropical regions insect abundance can be greatly reduced during winter (Fenton 1983), limiting the amount of food available to insectivorous microbats. A reduction in winter activity of bats in tropical regions also has been observed (Richards 1989) and several subtropical microbats frequently employ short torpor bouts coinciding with times of low T a, even during summer (Turbill et al. 2003; Jacobs et al. 2007; Vivier and van der Merwe 2007). As torpor use by insectivorous bats is rather flexible, and torpor bout duration generally increases with decreasing T a in thermo-conforming torpid individuals (Twente and Twente 1965; Avery 1985; Geiser and Kenagy 1988; Arlettaz et al. 2000; Turbill and Geiser 2008), it is possible that free-ranging tropical or subtropical microbats also use prolonged torpor during cold periods in winter.

Appropriate selection of hibernacula is important for microbats as they spend a large portion of their life within roosts and roost microclimates influence survival and other important life processes (Bronner et al. 1999; Kerth et al. 2001; Lourenço and Palmeirim 2004). However, roost selection is of critical importance particularly for tree roosting bats because they are more exposed to external environmental conditions and T a fluctuations than cave-dwelling bats (Genoud 1993). As T a affects torpor bout length and arousal frequency, roost choice will in turn influence torpor patterns and likely will also determine to some extent the use of prolonged torpor by microbats. Thus, differences in selected roost microclimate may explain differences in torpor patterns between individuals and even sexes (Ransome 1971; Humphries et al. 2002; Turbill and Geiser 2006; Jacobs et al. 2007).

The purpose of our study was to quantify the thermal biology, thermoregulatory behaviour and use of torpor of tree-roosting subtropical microbats under natural conditions during winter. We test the hypothesis that microbats use prolonged torpor even in a mild subtropical climate, especially at times when T a is low and foraging may be unproductive. Moreover, we predict that weather patterns and selection of roost microclimate will influence winter torpor patterns within and among individuals. To address these aims, we use temperature telemetry to locate roosts and continuously record the skin temperature (T skin) of free-ranging male and female Nyctophilus bifax (Vespertilionidae) living in a coastal, subtropical rainforest habitat of eastern Australia.

Methods

Our study was conducted in Iluka Nature Reserve (29°24′S, 153°22′E) on the north coast of New South Wales, Australia, during July and August 2007 (austral winter). This site is near the southern limit of the geographic distribution of N. bifax. Iluka Nature Reserve is subtropical according to the Köppen Climate Classification, as the mean T a of the warmest month is >22°C and the mean T a of the coldest month is between −3 and 18°C. The reserve contains an area of littoral rainforest and eucalypt forest, receives an average of 1,450 mm of rainfall annually, and the long-term (130 years) mean daily T a for July ranges from 9.7 to 19°C (Bureau of Meteorology, Australia). During our study period, T a, measured with temperature dataloggers (±0.5°C, iButton thermochron, Maxim Integrated Products, Inc., Sunnyvale, California, USA) under the canopy in the shade 2 m above the ground, ranged from 4 to 25°C.

Bats were caught in mist nets placed across pathways or in openings within the rainforest. Mist nets were opened just prior to sunset and kept open for several hours and checked every 15 min. Adult female (n = 4) and adult male (n = 4) N. bifax (mean body mass 10.1 ± 1.3 g, n = 8) were fitted with temperature-sensitive radio transmitters (~0.5 g, LB-2NT, Holohil Systems Inc., Carp, Ontario, Canada) providing data for each individual over periods of 10–27 days (preliminary data have been published in Stawski et al. 2008). After trimming a small patch of fur in the mid-dorsal region, we glued transmitters to the skin using a latex adhesive (SkinBond, Smith and Nephew United, Mount Waverley, NSW, Australia). Before attachment, transmitters were calibrated between 5 and 40°C in a water bath against a precision thermometer. One transmitter that was shed 28 days after attachment was re-calibrated and had drifted by <0.5°C over the entire temperature range from the original calibration. We used external transmitters owing to the difficulties of implanting transmitters into small bats and to maximise transmitter range. In small mammals, the difference between core T b and T skin is typically <2°C, particularly while the animal is resting or torpid (Audet and Thomas 1996; Barclay et al. 1996; Körtner and Geiser 2000a; Dausmann 2005). Further, T skin measurements are not simply measurements of T a, as during normothermic thermoregulation the T skin − T a differential increases substantially (Fig. 1a).

Fig. 1
figure 1

a T skin of a free-ranging female N. bifax (upper trace, dotted line) and T a (lower trace, smooth line) over 1 day showing that in normothermic bats T skin differs from T a. b T skin of a free-ranging male N. bifax (upper trace, dotted line) and T a (lower trace, smooth line) over 1 day showing differences between passive and active arousals. The black and white bars on each graph represent night and day, respectively

Full size image

Bats were released at the site of capture after being held for approximately 2 h. The following morning we tracked bats to their roost locations and did so on all days when they carried transmitters. A receiver/logger (Körtner and Geiser 1998) that recorded transmitter pulse rate and thus T skin every 10 min was set within range of each roosting bat. Receiver/loggers were checked each day to ensure they remained in signal range. We also took manual readings of pulse intervals several times each day to obtain additional data when bats were out of receiver/logger range and to ensure accuracy of logger readings. Data from loggers were downloaded every 2 to 5 days to a laptop computer. The exact location of several roosts was identified by observing bats leaving at dusk, but this was often not possible because of the thick vegetation and large number of possible roost sites in a small area. Approximate roost locations were marked with tape and recorded with a handheld global positioning system unit (GARMIN eTrex).

Torpor entry was defined as a rapid decrease (mean 0.1 ± 0.03°C min−1) in T skin to temperatures below 28°C. Active arousals were described by relatively fast increases (mean 0.5 ± 0.1°C min−1) in T skin to temperatures above 28°C, whereas passive increases in T skin to a maximum of 32°C were distinguished by slow rewarming rates that followed the general rewarming trend of T a (mean 0.1 ± 0.1°C min−1; Fig. 1b). Therefore, we calculated torpor bout duration from the time T skin fell below and rose above 28°C, with the exception of times when T skin briefly (<30 min) increased passively to temperatures up to 32°C without active rewarming. These passive arousals were included into torpor bouts.

Statistical analyses were conducted using Minitab Statistical Software (V 13.1). To determine whether the distribution of circular data, i.e. the timing of arousals, differed significantly from random we used the Rayleigh test (Zar 1999). Linear regressions were fitted using the method of least squares and compared using ANCOVA. Individuals were analysed separately and if slopes were the same they were pooled and regressed together. We report data as means ± SD for n is the number of individuals; N is the number of observations. To account for repeated measures in individuals we used the mean of the values for each individual when calculating means ± SD. We rejected the null hypothesis if the significance level was P < 0.05.

Results

Torpor and arousal patterns

All bats used torpor on each day of the field study. Patterns of torpor varied between warm periods when bats aroused each evening at dusk and cool periods when bats remained in prolonged torpor lasting up to 128.5 h (5.4 days; Fig. 2). All bats employed at least one torpor bout lasting >24 h. The mean duration of torpor bouts was 26.8 ± 11.3 h (n = 8, N = 114), although individuals showed considerable variation in mean torpor bout duration (Table 1). The mean duration of the longest bout recorded for each individual was 82.6 ± 28.2 h (n = 8). Overall, bats exhibited a total of 22 torpor bouts that lasted >24 h and 92 torpor bouts that lasted <24 h.

Fig. 2
figure 2

T skin of two free-ranging male (a, b) and two free-ranging female (c, d) N. bifax (upper trace, thick line) and T a (lower trace, thin line) over 10 days during July (males) and July/August (females) 2007. The dashed line at 28°C indicates the T skin torpor threshold

Full size image
Table 1 Mean ± SD for the given variables for individual free-ranging N. bifax during winter
Full size table

The log10 of the duration of torpor bouts was negatively correlated to mean T skin and T a of bats during each bout (P < 0.001, R 2 = 52.2%; P < 0.001, R 2 = 47.5%; Fig. 3a, b; respectively). The mean torpor bout duration (P = 0.61, T = −0.5, df = 5), the mean T skin during each torpor bout (P = 0.58, T = −0.6, df = 3), and the correlation between mean T skin/T a and torpor bout duration (ANCOVA, P > 0.05) did not differ between sexes.

Fig. 3
figure 3

Torpor bout duration (log 10) as a function of mean T skin during a torpor bout (a; P < 0.001, R 2 = 52.5%) and mean T a during a torpor bout (b; P < 0.001, R 2 = 47.5%) for four male (filled circles) and four female (open circles) N. bifax. When calculating mean T skin we included passive arousals above 28°C

Full size image

Times of arousal of torpid bats were strongly clumped (Rayleigh test, P < 0.001) with a main peak at around sunset (0.05 ± 0.1 h after sunset, n = 8, N = 92) and two smaller peaks, one shortly before dawn and another during the middle of the day (Fig. 4). Times of arousal did not differ between males and females (P = 0.44, T = −0.9, df = 3). When individuals aroused in the second half of the night, they had always first aroused at around sunset and re-entered torpor before arousing again near sunrise. Maximum daily T a affected a bat’s propensity to arouse in the evening and therefore the duration of torpor bouts (Fig. 5). On days when bats aroused, maximum T a was significantly higher (18.5 ± 0.8°C, n = 8, N = 100) than on days when bats remained torpid (16.3 ± 0.7°C, n = 8, N = 47, P < 0.001, T = 5.9, df = 13).

Fig. 4
figure 4

Times of arousals by N. bifax relative to the time of sunset (0 h). Filled bars represent males and open bars represent females. Each bar represents a 30-min period

Full size image
Fig. 5
figure 5

Mean maximum T a on days when bats aroused and on days when bats remained torpid

Full size image

Interbout normothermic periods lasted for a mean of 3.0 ± 0.8 h (n = 8, N = 96), with some variance among individuals (Table 1). The duration of these normothermic periods was positively correlated with mean nightly T a (P < 0.001, R 2 = 59.8%; Fig. 6a), but did not differ between the sexes (ANCOVA: slope P = 0.584, intercept P = 0.383). No correlation was observed between the duration of interbout normothermic periods and the duration of the preceding torpor bout (P = 0.292, R 2 = 1.2%; Fig. 6b).

Fig. 6
figure 6

Duration of interbout normothermia of male (filled circles) and female (open circles) N. bifax as a function of mean nightly T a (a; P < 0.001, R 2 = 59.8%) and the duration of the preceding torpor bout (b; P = 0.292, R 2 = 1.2%)

Full size image

Skin temperature

Daily passive fluctuations of T skin during prolonged torpor bouts (>24 h) varied between a minimum of 14.0 ± 2.5°C (n = 8, N = 47) and a maximum of 24.2 ± 2.6°C (n = 8, N = 47; mean daily range 10.2 ± 3.6°C, n = 8, N = 47; Table 1). The mean of the lowest minimum T skin recorded for each individual was 12.1 ± 2.5°C (n = 8), while the lowest individual minimum T skin recorded was 9.4°C (T a = 6.5°C) for a male and 10.8°C (T a = 7.5°C) for a female. The mean daily minimum T skin for all days, including torpor bouts of <24 h, was 14.7 ± 2.4°C (n = 8, N = 147) and varied among individuals (Table 1), but did not differ between sexes (P = 0.692, T = −0.4, df = 4).

Daily minimum T skin was positively correlated with the corresponding T a (i.e. measured at the same time as T skin) and also with the daily minimum T a (P < 0.001, R 2 = 33.6%; P < 0.001, R 2 = 40.2%; Fig. 7a, b; respectively). The relationship between daily minimum T skin and daily minimum T a did not differ between males and females (ANCOVA: slope P = 0.186; intercept P = 0.767). However, while the slope of this relationship was the same for all individuals, the intercept differed significantly among individuals (ANCOVA: slope P = 0.904; intercept P < 0.001), with minimum T skin ranging from 1.4 to 12.1°C above T a. During torpor bouts the time of daily maximum torpid T skin (25.2 ± 2.3°C, n = 8, N = 106) occurred 0.5 ± 0.4 h (n = 8, N = 106) after the daily maximum T a (17.4 ± 2.3°C, N = 29) and the time of daily maximum torpid T skin did not vary significantly between males and females (P = 0.249, T = −1.3, df = 5).

Fig. 7
figure 7

Daily minimum T skin of male (filled circles) and female (open circles) N. bifax as a function of corresponding T a (a; P < 0.001, R 2 = 33.6%) and daily minimum T a (b; P < 0.001, R 2 = 40.2%)

Full size image

Discussion

Our study shows that subtropical bats can employ prolonged torpor in winter. We observed multi-day torpor in all N. bifax with torpor bouts lasting up to 5.4 days, which is similar to, for example, torpor bouts of hibernating hamsters (Mesocricetus auratus) or chipmunks (Tamias amoenus) exposed to T a < 10°C (Pohl 1961; Geiser and Kenagy 1987). Our data refute previous claims that hibernation is restricted to temperate zone bats (McNab 1974) and that prolonged torpor in the subtropics is not possible because of the high prevailing T a (Henshaw 1970). We show that hibernation in bats is geographically widespread and that low and stable T a is not a prerequisite for prolonged torpor, although in subtropical regions hibernation patterns obviously differ from that in temperate climates because of the much milder conditions.

Although most work on torpor has been conducted in temperate regions, and it is widely believed that torpor is not required in warm climates, there is substantial evidence from field and laboratory work demonstrating that both mega- and microbats from the tropics and subtropics employ torpor (McManus and Nellis 1971; Genoud et al. 1990; Genoud 1993; Turbill et al. 2003; Jacobs et al. 2007; Kelm and von Helversen 2007; Vivier and van der Merwe 2007). Nevertheless, bats in all of these studies displayed short bouts of torpor lasting less than 1 day that are often described as “daily torpor”, in contrast to multi-day torpor in the hibernators. However, torpor bout duration is inversely related to temperature in thermo-conforming, torpid mammals and even in hibernators bouts of torpor lasting less than 1 day are common at T a > 20°C, especially when these animals are exposed to strong daily T a cycles (Twente and Twente 1965; Song et al. 1997; Körtner and Geiser 2000b; Geiser 2004). Although torpor bout duration may be similar at high T a in hibernators and daily heterotherms, torpor metabolic rates in hibernators are much lower than those of daily heterotherms (Song et al. 1997; Geiser and Brigham 2000; Geiser 2004) suggesting that torpor in hibernators and daily heterotherms differs functionally. In contrast to our study, torpor use in other tropical and subtropical bat species was examined for periods of <2 weeks. Our long-term study over 5 weeks, including variation in weather patterns, shows that N. bifax, like other subtropical bats, arouse from torpor daily during a period of warm weather in winter. However, this did not preclude entering prolonged torpor during cold spells. Thus, the use of short bouts of torpor by subtropical or tropical bats does not necessarily indicate the use of daily torpor exclusively (i.e. by a daily heterotherm), but perhaps more correctly represents short torpor bouts by a species capable of hibernation. It is likely that further work will identify prolonged torpor bouts in at least some other tropical and subtropical bats.

Torpor bout duration in N. bifax was strongly affected by T a as in previous studies on hibernating bats and rodents (Twente and Twente 1965; Ransome 1971; French 1982; Geiser and Kenagy 1987, 1988; Park et al. 2000; Rambaldini and Brigham 2008). Most torpor bouts observed here lasted for <1 day and likely occurred between nocturnal feeding periods made possible during mild nights in the subtropics when insects remain active. Even microbats from temperate climates are more active and may forage during warm weather in winter (Avery 1985; Hays et al. 1992; Arlettaz et al. 2000; Turbill 2008; Turbill and Geiser 2008). Nevertheless, the occurrence of prolonged torpor in our bats suggests that during cold periods it is not worthwhile for the bats to arouse and forage, likely because of the reduced abundance of flying insects on these nights (Richards 1989). We found that the average duration of prolonged torpor bouts was about three days and coincided with cold weather when average daily T a was cooler than ~12°C. Our results suggest that weather patterns, which determine thermoregulatory costs and insect activity, are an important predictor of torpor expression and activity patterns for subtropical bats during winter. Further, it seems possible that bats are able to sense changes in weather patterns via changes in barometric pressure (Turbill 2008; Turbill and Geiser 2008) and use this to decide whether to arouse or to remain torpid.

The T skin of torpid bats during most of our study was generally several degrees higher than T a and T skin passively followed the general trend of daily T a changes, suggesting that usually they were thermo-conforming as for many hibernating species (Lyman 1970). The large T skin − T a differential measured here is likely because T a was measured only 2 m above the ground, whereas bats generally roosted at heights above ~10 m. The lowest T skin recorded in our study for a male was 9.4°C (T a = 6.5°C) and for a female was 10.8°C (T a = 7.5°C) and daily minimum T skin matched day-to-day variation in external T a, hence the winter roosts of N. bifax typically provided little buffering against short-term fluctuations in external T a. However, different T skin during steady-state torpor suggest that roost microclimates differed among individuals and over time.

Unlike cave-roosting bats, tree-roosting N. bifax are exposed to large daily fluctuations in T a and consequently large T skin fluctuations during torpor. On some occasions, bats passively re-warmed to near normothermic levels without obvious active arousal. Experiencing such large daily fluctuations in T skin and hence T b may reduce the necessity for full periodic arousals from prolonged torpor bouts (Dausmann et al. 2005), but could also facilitate passive arousals with low energetic investment on relatively warm days when bats may attempt to forage (Ruczyński 2006; Turbill and Geiser 2008). In our study, maximum T skin during torpor bouts generally occurred shortly after maximum T a and on some days would remain high until around sunset when bats typically aroused. A passive increase of T b because of increasing ambient and roost T a reduces costs of rewarming substantially in comparison to active endothermic re-warming of T b (Lovegrove et al. 1999; Geiser et al. 2004; Turbill et al. 2008). Moreover, by experiencing daily fluctuations in T a, N. bifax can arouse in response to short term changes in winter weather patterns that are correlated with increased insect activity (Turbill 2008), but remain torpid on nights when the energy return would be insufficient.

Bats were normothermic for longer on warm than cool nights likely because of reduced thermoregulatory costs and increased insect food abundance (French 1982; Avery 1985; Hays et al. 1992; Park et al. 2000; Christie and Simpson 2006; Turbill and Geiser 2008). However, interbout normothermic periods lasted on average only about 3 h and were not influenced by the duration of the preceding torpor bout (Fig. 6b). This suggests that prolonged torpor in N. bifax, perhaps because of their small size (French 1982), does not require long recovery periods, at least for torpor bouts shorter than a week. Our findings raise the question of whether interbout normothermia is more than simply a post-torpor recovery event.

Most arousals in N. bifax occurred shortly after sunset, as in other studies of bats that can be relatively active during the hibernation season (Hays et al. 1992; Park et al. 2000; Turbill and Geiser 2008). However, on especially warm days some bats actively aroused around midday and showed two or more arousals throughout the night, often one just after sunset and another just before sunrise. This pattern of arousal and activity seems common during mild winter conditions, and was also observed, for example in Scotophilus microbats inhabiting a subtropical region (Jacobs et al. 2007). A possible explanation for arousals preceding sunrise is that, despite cool T a at this time, insect abundance seems to increase just prior to dawn. Presumably, on these nights a second opportunity for foraging just prior to the rest phase outweighs the relatively high costs for thermoregulation at this time.

Bats were more likely to arouse on warm nights than on cool nights throughout winter (Fig. 5). Similar results have been found in mild temperate regions, where hibernating bats are active and forage successfully on some winter nights that are warmer and less windy then usual (Avery 1985; Turbill 2008). The benefits of increased activity during winter, which may include opportunities to feed, mate or change roosts, in some circumstances evidently outweigh the costs of arousing from prolonged torpor (Christie and Simpson 2006). However, with a limited and fluctuating winter food supply, it seems important for microbats to select a roost T a that provides an accurate signal as to the probable costs and benefits of arousing and foraging during the hibernation season (Ransome 1971; Audet and Fenton 1988; Wojciechowski et al. 2007).

Our study did not reveal functional differences between males and females. However, we observed large individual differences that likely reflected selection of roosts with different microclimates. While our ability to exactly locate roosts was limited by the high roosting sites and dense, complex canopy vegetation, we established that N. bifax generally roosted under foliage or in large trees, likely with hollows, which is similar to a previous study (Lunney et al. 1995). These different roosts will have quite different microclimates and, as T a impacts the thermal biology of microbats, roost microclimate is likely important with regards to the energy reserves that are available to a microbat. Roost choice by some microbat species does appear to be more likely attributed to preferences of T a rather than of a particular roost (Bronner et al. 1999; Kerth et al. 2001). Further, mass and energy requirements of individual microbats vary, which likely affects the way they use torpor to optimise their energy reserves and therefore their roost T a requirements. In some microbat species individuals will often roost in different microclimatic conditions even when roosting in the same area (Boyles et al. 2007). Therefore, it seems likely that hibernaculum choice plays an important role in the expression of torpor patterns and may be the reason for some of the differences in torpor use seen among individuals in our study.

In conclusion, we have shown that the use of prolonged torpor in bats is not restricted to regions that experience cold and stable temperatures, but can be used in climates that experience relatively high and variable temperatures. In a subtropical climate during winter N. bifax was able to enter torpor for prolonged periods despite experiencing large daily changes in T a and T skin. However, N. bifax were also able to take advantage of warmer periods during winter to arouse and likely to feed. Not only do patterns and use of torpor vary greatly among species and climates, but they also appear to differ among individuals. These individual differences are largely a result of different roost locations, which could be attributed to varying energy requirements. Therefore, torpor patterns in microbats appear to be highly flexible and many species are likely to be opportunistic in their use of torpor.