Seasonality of torpor patterns and physiological variables of a free-ranging subtropical bat

Seasonal variations in climate occur in all habitats, with the most extreme seasonal changes generally occurring in temperate or arctic regions. However, weather conditions in tropical and subtropical regions do vary seasonally and these changes can also be substantial. Changes in ambient temperature (T_a) and rainfall patterns in subtropical regions from summer to winter can be pronounced, resulting in different primary productivity and consequently different physiological demands on animals. Insectivorous bats are small (most weigh <25 g) and have a disproportionately large surface area to volume ratio. Consequently, they need large amounts of energy in comparison to larger animals to remain at normothermic body temperatures (T_b) at T_a lower than their thermal neutral zone (TNZ) and as a result may also be more affected by seasonal changes in weather than other mammals (Brown, 1999; Turbill et al., 2003; Vivier and Van Der Merwe, 2007). Thermoregulatory energy demands would require that insectivorous bats consume enormous amounts of food on cool nights to remain normothermic. However, many moth species and other insects are unable to remain active at low T_a, which greatly reduces the amount of food available to insectivorous bats on cool nights (Taylor, 1963; Richards, 1989; Turbill et al., 2003). Consequently, insect abundance can significantly change daily and also seasonally (Hickey and Fenton, 1996; Jacobs et al., 2007; Stawski and Geiser, 2010). To deal with such energetic constraints and to conserve energy, many insectivorous bats employ torpor, which is characterised by pronounced reductions in T_b, metabolic rate (MR) and water loss (Hock, 1951). Moreover, the use of torpor is often initiated by a decrease in food availability and/or T_a (Wang, 1989), suggesting that it would be beneficial for insectivorous bats to be flexible in their use of torpor to suit current seasonal and daily weather conditions and therefore insect abundance.

Studies undertaken on the seasonal activity of insectivorous bats from a variety of habitats have found that activity is generally highest during summer and lowest during winter (Richards, 1989; Brigham and Geiser, 1998). However, laboratory studies on torpor use in cool-temperate insectivorous bats established that season per se did not have a pronounced effect on torpor use and physiological variables in comparison to environmental conditions and food availability (Geiser and Brigham, 2000; Wojciechowski et al., 2007). Surprisingly, in the subtropical nectarivorous blossom bat (Syconycteris australis) seasonal functional changes were observed, but these were the opposite of those commonly observed in other mammals because torpor bouts were longer and T_b lower during summer than in winter (Coburn and Geiser, 1998). However, S. australis primarily feed on nectar that is more abundant during winter than during summer, again suggesting that torpor use in this species is predominantly determined by food availability. Importantly, all seasonal functional studies on torpor patterns were undertaken in the laboratory. Seasonal data in free-ranging subtropical insectivorous bats are currently restricted to descriptive information on the seasonal differences in torpor use (Vivier and Van Der Merwe, 2007), whereas information on seasonal changes in physiological variables of torpor patterns and their systematic analysis are entirely lacking.

Therefore, the aim of our study was to obtain detailed quantitative data on the thermal biology of the insectivorous northern long-eared bat, Nyctophilus bifax Thomas 1915, during summer and winter from subtropical Australia by using temperature telemetry. Little biological and ecological information is available about N. bifax. However, it is known that they are a tree-roosting species, give birth to young during October, lactate from November to December, and feed on moths, ants and click beetles (Churchill, 1998). We predicted that N. bifax will use torpor regularly throughout the year in a subtropical habitat, especially during adverse weather conditions and low food availability. Furthermore, considering the substantial seasonal changes in weather and food availability (Stawski and Geiser, 2010) we hypothesised that thermal biology and torpor patterns of free-ranging N. bifax differ seasonally.

Data on the seasonal patterns and physiological variables of torpor use in free-ranging subtropical insectivorous bats, Nyctophilus bifax, were collected at Iluka Nature Reserve (29°24′S, 153°22′E) in Australia, during the Austral winter (July to August 2007) and summer (February to March 2008). Iluka Nature Reserve is a subtropical area of littoral rainforest and eucalypt forest. Some data from our winter study have been published in a descriptive study (Stawski et al., 2009) and some data from our summer study have been published to test the cost-benefit hypothesis of torpor use (Stawski and Geiser, 2010), but all data were re-analysed and new data were included for this seasonal comparison.

We report data from eight non-reproductive individual N. bifax from winter (measured for 10–27 days) and thirteen non-reproductive individual N. bifax from summer (measured for 1–12 days). Bats were captured with mist nets and were weighed using a pro-Fit™ pocket/travel scale (0.1 g resolution, InterTAN Australia PTY Limited; Cat. No. 63-9534). Although N. bifax were non-reproductive during the times of our study, females were checked to confirm that they were not lactating or pregnant. After removal of a small patch of fur, temperature-sensitive radio transmitters (∼0.5 g, LB-2NT, Holohil Systems Inc., Carp, ON, Canada) were attached to the mid-dorsal skin region using latex adhesive (SkinBond; Smith and Nephew United; Mount Waverley, NSW, Australia) to obtain data on the skin temperature (T_skin) of N. bifax. Owing to the difficultly of implanting transmitters into small microbats and because the difference between T_b and T_skin for small mammals while torpid is <2.0°C (Audet and Thomas, 1996; Barclay et al., 1996), we felt that the use of external transmitters was appropriate for our study. To ensure the accuracy of T_skin data, transmitter pulse rate was regressed against transmitter temperature, by calibrating transmitters before use at temperatures between 5.0 and 40.0°C in a water bath against a precision thermometer (0.1°C resolution). We found that transmitters drifted by <1.0°C over the entire temperature range as we were able to re-calibrate three transmitters 7 to 28 days after initial calibration, which were worn by bats and shed.

Bats were located on every day they carried transmitters and the T_skin of each bat was monitored every day via telemetry during both study periods. We determined the exact location of several roosts, when possible, during summer and winter, however, this was often not feasible 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). A remote receiver/logger (Körtner and Geiser, 2000a) was used to continuously record the T_skin of each bat once every 10 min when bats were in transmission range. Every 2–4 days we downloaded the data from the receiver/loggers to a laptop computer. T_a in both summer and winter was measured with temperature data loggers (±0.5°C, iButton Thermochron DS1921G, Maxim Integrated Products Inc., Sunnyvale, CA, USA) that were placed in the same location during both seasons under the canopy of the rainforest in the shade 2 m above the ground.

During the summer study when T_skin was <28.0°C for >30 min bats were considered to be torpid. We felt that the use of this definition for torpor was appropriate, as the T_b–T_skin differential of torpid small mammals is typically<2.0°C and many studies use T_b<30.0°C to define torpor (Barclay et al., 2001). For our winter study, torpor bouts were similarly defined, with the exception of passive fluctuations where T_skin occasionally increased to temperatures >28.0°C (max 32.0°C) for <30 min without an obvious active arousal and such passive fluctuations were included into torpor bouts.

Data are reported as means ± s.d. for the number of individuals (n) and the number of observations (N). StatistiXL (V 1.8, 2007) was used to conduct most statistical analyses. A Rayleigh test was used to determine whether circular data (timing of arousals and torpor entries) differed significantly from random. To account for repeated measures in individuals when calculating means ± s.d. for torpor bouts and T_skin, the mean of the values for each individual were used. For regressions and timing of events, such as arousals and torpor entries, multiple data points for each individual were included in analyses. If the significance level was P<0.05, the null hypothesis was rejected. To determine the random effect of individuals we compared linear regressions in StatistiXL to check for differences in slopes by using ANOVA. No differences were found in the slopes of individuals for any of the regressions and therefore individuals were pooled and regressed together. Minitab (V 13.1, 2000) was used to determine the effect of body mass on all statistical analyses. Body mass and season were used as covariates in a general linear model for mean torpor bout duration, mean maximum torpor bout duration, mean minimum T_skin and minimum T_skin–T_a and it was found that body mass had no effect on any of these variables, whereas season did have a significant effect on all variables. For all regressions for both seasons body mass was included as a predictor to determine the effect it had on the given variable (torpor bout duration, T_skin, and T_skin–T_a). Body mass had no effect on any of the variables for both seasons, except for T_skin during summer and this is included in the results.

T_a during winter varied between a minimum of 8.2±2.2°C (n=29) and maximum of 17.4±2.3°C (n=29), with a mean daily range of 9.2±1.9°C (n=29). During summer, T_a varied between a minimum of 18.3±1.2°C (n=26) and a maximum of 25.0±2.3°C (n=26), with a mean daily range of 6.7±2.2°C (n=26). As predicted, both minimum and maximum T_a during winter were significantly lower than during summer (P<0.001, t=21.3, d.f.=44.8; P<0.001, t=12.3, d.f.=53; respectively), however, the daily range of T_a was significantly larger during winter than in summer (P<0.001, t=4.6, d.f.=53; Fig. 1). Mean daily T_a was higher during summer (21.2±1.3°C, n=26) than in winter (12.4±2.2°C, n=29; P<0.001, t=18.7, d.f.=46.3; Fig. 1). During winter, mean night T_a was significantly lower (10.9±2.4°C, n=28) than during summer (20.1±1.2°C, n=25; P<0.001, t=17.6, d.f.=40.4; Fig. 1).

Torpor was used by all individuals on all days (100%) during winter and on 85% of observation days during summer. Torpor bout durations ranged from 0.8 to 128.5 h in winter and from 0.7 to 21.2 h in summer. Mean torpor bout duration was significantly longer during winter (26.8±11.3 h, n=8, N=114) than in summer (3.2±1.3 h, n=13, N=122; P<0.001, t=6.0, d.f.=7.1; Table 1) and so were mean maximum torpor bout durations recorded for each individual (winter: 83.7±27.6 h, n=8; summer: 8.2±6.2 h, n=13; P<0.001, t=7.6, d.f.=7.4). Daily patterns of torpor during summer varied (Fig. 2A) and the most common pattern of torpor was two or more torpor bouts during the day (55%); less common were one morning or one afternoon torpor bout (21.7%), remaining torpid during the day (8.7%), and lastly, no torpor (14.6%). Winter torpor patterns generally involved short normothermic periods in the evening or no normothermic periods over several days (Fig. 2B). Torpor bout duration (log₁₀) decreased with increasing T_a during winter (F_1,107=96.0, P<0.001, R²=0.5), however, this relationship was not significant during summer (P=0.2; Fig. 3). Nevertheless, when torpor bout duration (log₁₀) and T_a from summer and winter were pooled and regressed together this relationship was significant (F_1,226=410.3, P<0.001, R²=0.6) and increased R² in comparison to winter-only data. The percentage of bats that used torpor on a given day during summer increased with decreasing daily minimum T_a (F_1,22=4.9, P=0.04, R²=0.2), average T_a (F_1,22=6.9, P=0.02, R²=0.3) and evening (sunset to midnight) T_a (F_1,21=16.5, P<0.001, R²=0.5).

Daily entries into torpor during summer reached an absolute peak at 36.0±26.6 min (n=12, N=42) after sunrise, but this pattern was not statistically significant (Rayleigh Z=2.0, P=0.132; Fig. 4A). Timing of arousals during summer was non-randomly distributed (Rayleigh Z=7.1, P=0.001; Fig. 4A) with a mean time (angle) of 13:05±4:44 h (n=13, N=128) and arousals for possible foraging periods occurred 14.5±21.4 min (n=13, N=41) before sunset. During winter, both entries into torpor and arousals from torpor were non-randomly distributed (Rayleigh Z=24.6 and 72.0, P<0.001; Fig. 4B) at a mean time (angle) of 19:56±4:00 h (n=8, N=122) and 16:26±2:29 h (n=8, N=116), respectively. Arousals for probable foraging periods during winter occurred 4.5±11.4 min after sunset (n=8, N=92).

The lowest individual T_skin recorded was 9.4°C during winter and 16.0°C during summer. Mean minimum torpid T_skin was significantly lower during winter (14.7±2.4°C, n=8, N=147) than during summer (23.1±2.3°C, n=13, N=63; P<0.001, t=7.8, d.f.=19.0; Fig. 1, Table 1). During summer, daily minimum torpid T_skin occurred 196±191 min (n=13, N=57) and during winter 197±99 min (n=8, N=144) after the daily minimum T_a. Minimum torpid T_skin was positively correlated with the corresponding T_a during winter (F_1,143=71.8, P<0.001, R²=0.3) and also during summer (F_1,56=41.9, P<0.001, R²=0.4; Fig. 5A). However, the slope of this relationship differed significantly between seasons (summer=1.3; winter=0.6; F_1,198=10.7, P<0.001). Furthermore, it was found that body mass had a significant effect on T_skin during summer (P=0.005) and changed the slope of the relationship between minimum torpid T_skin and T_a (new slope=1.2). Partial regression plots for the adjusted relationship between minimum torpid T_skin and T_a have been included for both summer and winter (Fig. 5B). The differential between minimum T_skin and the corresponding T_a was larger during winter (5.4±1.5°C, n=8, N=140) than during summer (2.8±1.8°C, n=13, N=57; P=0.003, t=3.4, d.f.=19.0). Further, the relationship between this differential and T_a was significant during winter (F_1,143=39.3, P<0.001, R²=0.2), but not during summer (P=0.2; Fig. 6).

We provide the first evidence of seasonal changes in torpor patterns and thermal physiology of a free-ranging insectivorous bat from a subtropical region. Our study shows that torpor in winter is more frequent, deeper and longer than in summer. Although short bouts of torpor were observed during both seasons, the range of the duration of torpor bouts during winter (0.8 to 128.5 h) was much greater than during summer (0.7 to 21.2 h). Consequently, the longer torpor bouts during winter suggest that N. bifax are more energetically constrained during winter than in summer. Interestingly, in addition to the predicted lower T_a in winter, it was found that the daily range of T_a experienced by bats studied here during winter was significantly greater than during summer. These seasonal differences in T_a will affect insect abundance and it was found that insect abundance at Iluka Nature Reserve during summer is 14-fold of that during winter (Stawski and Geiser, 2010). Therefore, T_a and insect abundance are likely responsible for much of the seasonal changes in torpor patterns of N. bifax in this subtropical climate.

As in other species of insectivorous bats (Ransome, 1971; Park et al., 2000; Rambaldini and Brigham, 2008), torpor bout duration in N. bifax decreased with increasing T_a during winter, but this relationship was not significant during summer. However, the weak effect of T_a on torpor bout duration during summer was most probably due to the narrow range of T_a experienced, but when this range was increased by adding all data from winter, the relationship became statistically significant. This suggests that there is no clear seasonal change on the influence of T_a on torpor bout duration. Even though T_a showed no effect on torpor bout duration during summer it was found that T_a does affect the percentage of bats that enter torpor on a given day. Therefore, although the weather was most probably mild enough during summer to forage and insect abundance was high, torpor was still used more often on cooler days during summer for energy conservation.

Preliminary data for MR of N. bifax (C.S., unpublished data) suggest that they are similar to those of N. geoffroyi. Therefore, by using resting and torpid MR of N. geoffroyi (Geiser and Brigham, 2000), we were able to calculate approximate energy expenditure for N. bifax during summer and winter. Normothermic thermoregulation of N. bifax for 1 h during the day in summer at T_a 25°C (near mean daily max T_a) requires 532 J/h, whereas in winter at T_a 15°C (near mean daily max T_a) they require 1165 J h^–1, more than twice as much. However, torpor at T_a 15°C during winter reduces energy expenditure to approximately 16 J h^–1, which is a mere 1% of the resting MR for the same time period. The torpid MR during summer (T_a 25°C) at 39 J h^–1 is also low, only 7% of the energy used during normothermia. Therefore, it seems worthwhile for individuals of N. bifax to use torpor as observed here regularly throughout the rest period in both summer and winter. However, in comparison to winter, arousal from torpor and remaining normothermic during summer requires much less energy because of the relatively high T_b, so they can be more flexible with torpor use as shown by the frequent arousals.

Although entries into torpor during summer in the current study occurred statistically randomly throughout the day, many bats entered torpor shortly before or after sunrise similar to temperate insectivorous bats (Turbill et al., 2003; Rambaldini and Brigham, 2008). By contrast, during winter N. bifax entered torpor much earlier during the night if they had aroused in the evening, probably due to the cold night T_a. Arousals from torpor for possible foraging trips during both summer and winter occurred near sunset, as for other insectivorous bats (Park et al., 2000; Turbill et al., 2003; Turbill, 2006). However, whereas most arousals for possible foraging trips during summer occurred 15–30 min before sunset, during winter most occurred just shortly after sunset. Frequent daily arousals (two to three times per day) from torpor during summer, as observed here, have been described in another subtropical microbat species (Vivier and Van Der Merwe, 2007) and are probably because costs of arousals are low for small mammals as they have only little tissue to heat, they often passively arouse to a certain T_b and because of the above mentioned relatively high T_b during summer (Geiser and Baudinette, 1990; Turbill et al., 2008; Warnecke et al., 2008).

During both seasons T_skin in N. bifax was generally a few degrees above T_a and time of daily minimum torpid T_skin was around 3 h after the time of daily minimum T_a, which is most likely a reflection of ambient conditions within roosts. Also, during summer, the delay in minimum torpid T_skin in comparison to minimum T_a may be a result of bats foraging or being normothermic during the part of the night when T_a is at its lowest, as they often entered torpor after sunrise. Daily minimum torpid T_skin during winter was significantly lower in comparison to summer, and was positively correlated with T_a during both winter and summer, similar to other Nyctophilus species (Morris et al., 1994; Geiser and Brigham, 2000). However, this relationship differed between seasons and the slope was greater during summer (1.3) than in winter (0.6). Furthermore, it was found that body mass influenced daily minimum torpid T_skin during summer but not during winter, which may contribute to the seasonal variation in the relationship between daily minimum torpid T_skin and T_a. It is important to note, however, that the influence of body mass on the relationship between daily minimum torpid T_skin and T_a was weak. The difference in slopes between seasons may also be explained by T_skin–T_a, which was significantly higher during winter than in summer. Additionally, the relationship between the T_skin-T_a differential and T_a was significant during winter, but not during summer. Throughout winter T_skin–T_a decreased as T_a increased, suggesting that as T_a becomes lower the bats either begin to thermoregulate or are residing in roosts that remain above a certain T_a. During summer the T_skin–T_a differential remained relatively constant over the range of T_a experienced and the smaller differential may be a reflection of the higher T_a during summer, which would allow the bats to continuously thermo-conform and lower their T_skin to near T_a. This in turn may explain why the relationship between daily minimum torpid T_skin and T_a is stronger during summer than in winter.

The seasonal differences in the relationship between T_skin and T_a may also be explained by roost choice, such that N. bifax may be choosing more thermally insulated roosts during winter. As T_a was measured in only one location during both studies we do not know how different roost T_a was to external T_a. However, it is known that N. bifax uses a wide variety of roosts, such as foliage and tree hollows (Lunney et al., 1995). Previous studies have shown that roost choice influences torpor bout durations, T_skin and also the amount of energy expended while roosting (Willis and Brigham, 2005; Stawski et al., 2008). It is therefore likely that individuals select specific roosts depending on their energy requirements, the current weather conditions and also the season (Boyles et al., 2007).

Evidence from some studies on bats suggests that torpor use may be more dependent on food availability and environmental variables rather than season per se (Coburn and Geiser, 1998; Wojciechowski et al., 2007). The seasonality of torpor use seen in many species is therefore most probably a response to changes in food availability as food abundance is often seasonal and is affected by ambient conditions. For example, activity by insects is dependent on T_a and many insects are unable to fly or move at low T_a, resulting in a decline in insect abundance from summer to winter (Richards, 1989; Stawski and Geiser, 2010). Also, many plant species flower seasonally because of rainfall patterns and other weather conditions, which explains why some nectarivorous mammal species use torpor more often during summer than winter (Coburn and Geiser, 1998; Boyer and Barnes, 1999). Therefore, the effect of T_a in some circumstances may be indirect, as it is more likely for an individual to become torpid when energetically compromised by lack of food. Our data support this because torpor use is more prevalent during winter, when insect abundance was lowest (Stawski and Geiser, 2010). However, from an energetic point of view it is still beneficial for a small animal to remain torpid while not active during any time of the year to conserve energy, regardless of food abundance.

In the past it was assumed that low and stable T_a is necessary for an animal to hibernate (Henshaw, 1970). In the subtropical winter even when T_a was variable and much warmer than in temperate winters, N. bifax entered prolonged torpor bouts (up to 128.5 h) regularly which clearly classify them as ‘hibernators’ (Geiser and Ruf, 1995; Schmid and Ganzhorn, 2009). Consequently, the short bouts of torpor observed in our study during summer, as found for dormice, most likely are short bouts of hibernation, which simply are shortened because of high and fluctuating T_a, rather than daily torpor (Geiser, 2004; Bieber and Ruf, 2009). This suggests that N. bifax, like N. geoffroyi, is reducing MR and T_b significantly more during these short bouts of hibernation in comparison to the shallow torpor bouts seen in daily heterotherms, thus maximising energy savings (Geiser and Brigham, 2000; Geiser, 2004). Our current study and preliminary data that show MR are similar in both seasons (C. S., unpublished data) provide further support for the argument that, apart from the observed temperature effects, there is no apparent functional difference between the physiology of prolonged and short bouts of torpor of hibernators, which are capable of showing both (Hock, 1951; Geiser and Brigham, 2000; Bieber and Ruf, 2009). N. bifax, like many other species of bats and other mammals, appear to be ‘opportunistic hibernators’ and make use of increased food availability during occasional increases in T_a during winter and enter prolonged bouts of torpor when conditions are unfavourable (Körtner and Geiser, 2000b).

To conclude, it seems that the seasonal differences seen in torpor use is to a large extent a result of seasonal changes in food abundance as caused by changes in T_a and an individual's energy needs. This further suggests that species that use prolonged torpor during winter and also short bouts of torpor during summer are functionally using the same type of torpor, and that differences in torpor patterns and thermal physiology are mainly ecological rather than physiological. Lastly, it seems clear that N. bifax is a hibernating tropical/subtropical species that use torpor regularly throughout the year during times of energetic stress and even during times when they are in good body condition and food is abundant, probably to minimise exposure to predators (Stawski and Geiser, 2010).

We would like to thank Stuart Cairns, Gerhard Körtner, Brad Law, Anaïs LeBot, Alexander Riek, Margaret and Michal Stawski, Christopher Turbill, Courtney Waugh and Philip Withers for their contributions to this study. Permits to undertake this research were issued by New South Wales National Parks and Wildlife Service and the Animal Ethics Committee of the University of New England.