During the den entry period, Tb and activity level appeared to be influenced by environmental factors, such as TA and snow depth. TA declined before activity levels and both parameters preceded the decline in Tb and the first snow event (HR and the SDANN both declined later than these variables). The den emergence process began with increases in TA, Tb, HR, SDANN, and activity level, in that order, apparently independently of snow depth. The observation that SDANN increased only 13 days after the increase in HR (i.e. a gradual shift in autonomous nervous system balance) suggested that the first increase in HR was adaptive thermoregulation (through a temperature-dependent increase in metabolic activity). Although this is well described in small hibernators and even in humans during induced therapeutic hypothermia, this observation would require further investigations.
The BCPA greatly increased our ability to estimate den entry and exit dates (Additional file 1: Figure S4). A simple evaluation of GPS positions would not have sufficed, because the transition period between hibernating and active states was drawn out, and some bears regularly or periodically returned to the denning area after emergence. We observed significant differences in the timing of den entry and exit between years, with the latest den entry dates during the warmer years. As climate change projections predict warming wintertime temperatures for the [23], shorter hibernation periods can be expected. For example, the decline in Tb before den entry was significantly correlated with TA, whereas HR was independent of the predictor variables we tested, except for day of the year. This suggests that environmental variability can affect behavioral and physiological aspects of hibernation independently in the bear.
Our results regarding timing of den entry and exit were consistent with a previous study on this population [24]; den entry and den exit dates occurred within the same ranges and males and older animals entered the dens later and exited earlier than females and younger animals [25]. Although a study on brown bears in Alaska documented a correlation between den emergence and the timing of snow melt [26], the bears in our study area had emerged from their dens either before or during snow melt (Fig. 3). Moreover, during emergence, there was no apparent impact of snow depth or photoperiod on Tb or HR; however, activity level was affected by both snow depth and photoperiod. This result indicated that, although snow depth was associated with a change in the behavior of the bears (e.g. den entry, consistent with the previous report [25]), it was not associated with any physiological change measured in this study. However, we found a difference in den entry from year to year, with bears entering the dens earlier in a colder year.
Drivers of den entry
Although Tb was correlated with several factors during den entry (Additional file 1: Table S2), convergent cross mapping revealed the greatest causation due to SDANN, followed by TA (Table 1). SDANN and HR were closely related at this stage of hibernation, likely because activity was low, eliminating activity’s confounding effect on HR. SDANN declined steeply just before den entry. Although the SDANN is not the best proxy to assess autonomic nervous system balance, we were unfortunately not able to use another index due to the duration of the experiment and the storage capacity required for a full ECG to be stored. Therefore, it is difficult to determine if changes in SDANN should be attributed to changes in the parasympathetic or sympathetic nervous system or both. Previous studies [27] have shown that hibernating bears have an enhanced respiratory sinus arrhythmia, indicating increased parasympathetic nervous system activity. Based on the literature on small hibernating mammals (e.g. [28, 29], the observed decrease in SDANN likely indicated that a predominant parasympathetic tone drives metabolic suppression and decreases in most physiological functions. In small hibernators, cooling is achieved by both metabolic depression and passive body cooling. Our data are in the line with the previously reported observation that large hibernators initially rely on metabolic depression to achieve depressed metabolic rate during hibernation [9, 30] to a larger extent than passive body cooling, although both mechanisms occur. The early rise in Tb in the bear is in contrast to most hibernators, with likely two involved mechanisms that cannot be dissociated but likely happen sequentially: Q10 effect due to slow warming with concomitant ANS activity and final warming from a massive SNA burst.
Both SDANN and HR stabilized 20 days after den entry, but it took an additional 10 days for Tb to stabilize, probably because of the bear’s large body mass and decreased heat exchange in the den. This is similar in sequence to that found in woodchucks (Marmota monax), where the decrease in metabolic rate occurred in 6 h, whereas the drop in Tb continued for 12 h [31], and in golden hamsters (Mesocricetus auratus), with metabolic rate decreasing for 3 h and Tb dropping for 8 h [32, 33]. However, the mechanisms of metabolic rate reduction are thought to differ between large and small hibernators [9]. Large hibernators, such as bears, are expected to rely to a greater extent on active metabolic suppression to reduce their metabolism, due to their larger body size, compared to small hibernating species, which benefit more from the Q10 effect in torpor. Even small marsupials (<20 g) actively suppress their metabolism during torpor [9]. Using active metabolic suppression, bears are able to reach metabolic rates (despite having a Tb above 30 °C) as low as small hibernators in deep (<5 °C Tb) torpor [34].
Including SDANN in our study proved to be particularly valuable. In contrast to previous approaches [33], HR was not used to infer metabolic rate, because, this parameter is confounded by activity and stress [33] which are expected prior to hibernation with the combination of both the hunting season and den entry behavior [13]. However, activity affects SDANN to a much lesser extent [22]. By including SDANN, we avoided the confounding effect of activity on HR and had indirect access to information on the sympathovagal regulation of metabolism. One study in dogs found that HRV did not differ between slow movements, lying, sitting, or standing, but did change when a favorite toy was presented [22]. In stressful situations, dogs had consistently increased HR and decreased HRV [35]. Therefore, whereas HRV may often be connected to HR, it is more independent of movement-induced changes. Variability in HR can be caused by changes in thermoregulation, circadian rhythms, respiration, blood pressure, and both physiological and psychological stressors and can be used to evaluate the state of balance between the sympathetic and parasympathetic nervous systems [36], but does not give an overall level of either system’s activity.
Tb in captive brown bears was reported to decline gradually over 5 weeks from the date that food and water were removed [37]. Our finding that changes in Tb began long before changes in HR suggested that previous studies focusing on captive bears with an artificially defined end of the food/water season might not represent the actual sequence of events in the wild. In our study, SDANN declined steeply just before den entry. This change was probably associated with the enhanced respiratory sinus arrhythmia previously reported to occur in hibernating bears [27]. Based on the literature on small hibernating mammals, the observed decrease in SDANN suggested that a massive parasympathetic tone, likely with a reduction in sympathetic activity, drives metabolic suppression and decreases in most physiological functions. The role of SNS in thermogenesis and cardiovascular control has been the topic of a number of experiments starting more than 60 years ago [38]. In studies of small hibernators, the initial fall in HR during entry into hibernation is due to parasympathetic activation and the exit due to SNS activation [39]. Treatment with atropine, an inhibitor of parasympathetic pathways, prevented 13-lined ground squirrels (Citellus tridecemlineatus) from entering hibernation [40]. Our results are in line with those from previous studies and suggest that increased parasympathetic activation plays a key role in the reduction in HR at den entry in bears as well, but does not rule out potential decreases in sympathetic nervous system activity.
Drivers of den exit
Although den exit was not correlated with either TA or photoperiod, the bears exited the dens at TA of 3.7° ± 1.3 °C. A bear den is not an adiabatic shell, however, the inside air temperature could easily rise, depending on the type of den (ant hill, under rocks or nests [41]). The fairly narrow range of Tb between bears on the day of exit (36.7 ± 0.15 °C, Additional file 1: Table S1) suggested that the bears exited when they reached a specific set point. At TA of > 0 °C, water also could start draining into the den, causing the bear to become uncomfortable and leave the den. American black bears (Ursus americanus) in artificial dens have a mean lower critical TA of 5 °C, below which the bears’ thermal conductance increased [42]. This suggests that the bears’ cue to exit the den was that they became too warm when the temperatures rose in springtime or that they were seeking more optimal temperatures outside the dens.
That TA does not drive Tb during the phase before exit (period 5, Table 1) might be due to the adaptive thermoregulation that occurred over several months, making the TA immediately around the day of exit less important. It could also be that the den temperature was more relevant, as the bears exited when TA reached approximately 3.7 ± 1.3 °C (Additional file 1: Figure S1), nearing the lower critical temperature for established for black bears and polar bear (Ursus maritimus) cubs [42, 43], possibly because the den temperature was above thermoneutrality.
Tb started rising 2 months prior to exit, whereas HR rose a month later, and was followed by SDANN (20 days prior to exit) and activity (10 days prior to exit). During the first part of arousal, the causation analysis revealed that TA caused Tb and Tb caused HR. There was no causal relationship between SDANN and Tb, but a different trend emerged 20 days before exit.
The gradually decreasing difference between Tb and TA during the first period (Additional file 1: Figure S5), suggests that bears were thermoregulating at a lower thermoregulatory set point during hibernation. This is consistent with recent findings from captive bears showing a negative relationship between den temperatures and hibernating metabolic rates [42]. Then, SDANN started rising and may have caused Tb and HR to rise, likely via an increase in sympathetic nervous system activity, a decrease in parasympathetic nervous system activity, or a combination. At this stage Tb lost its causal association with TA although Tb-TA remained stable during the second period, suggesting that euthermic metabolism was reestablished later by active thermogenesis, likely involving the sympathetic nervous system. The exact roles of the sympathetic and parasympathetic nervous systems in this process can only be assessed by direct measurements of sympathetic and parasympathetic nervous system activity in free-ranging conditions, which would be difficult to conduct.
This second phase of den exit was driven by SDANN, with SDANN driving Tb (p <0. 01). When SDANN began to rise, the thermoregulatory pattern shifted. This could indicate transitioning out of hibernation i.e., sympathetic nervous system activation combined with potentially a more profound change in metabolic state [42]. Activity increased from 10 days before exit, showing a causative relationship with the increase in Tb (activity drives Tb, p=0.05). Den exit occurred when Tb was almost at euthermia (mean 36.7 °C), nearing the lower critical temperature for bears [42]. Tb and SDANN stabilized quickly (within two weeks after den exit), but HR and activity took longer, indicating that the bears took longer to return to their original activity levels.
Although shivering may play a role in active thermogenesis, it occurs at the end of arousal in the species studied to date, excluding tropical hibernators [44]. Increased Tb allows restoration of enzyme functioning through a Q10 effect and contributes to restoration of muscle function. Early on, the processes start with SNS activation of the vascular system to increase body temperature and heart rate [44]. The role of SNS in thermogenesis in addition to vascular control has also been the topic of numerous investigations starting from the early studies of Lyman. Studies on American black bear in the laboratory show a role for shivering at the end of arousal [34], although our results show that it was less important in free-living conditions.
A recent study on captive American black bears found that metabolic rate was related to den temperature and showed that larger bears showed more variation in length of Tb cycles [42]. During experimental manipulations of den temperature, they found no direct relationship between den temperature and Tb, although the time between peaks in Tb became longer at higher den temperatures. The authors suggested, based on a single bear that increased its Tb to 35.9 °C when the den warmed to 10 °C, that the bear may have inhibited heat dissipation mechanisms. It is not clear whether this was merely an effect of being inside an isolated den or was a physiological phenomenon. In addition, they found that the lower critical temperature varied from 1° to 10 °C, from which they concluded that the smaller bears partially compensated for higher thermal conductance with increased metabolic rate. Interestingly, they found no relationship between TA and den temperatures. Although it was not possible to measure the den temperatures in our study, we would expect a correlation with TA, because the bears in our study were not in adiabatic shells; they were under rocks or tree roots or in anthills, with oxygen exchange varying from a small ventilation hole to large openings under rocks. We found, however, that TA drives Tb during the first phase, and the differential between Tb and TA decreased until the point in the spring when HRV rose. Although [42] found a negative relationship between TA and metabolic rate, we conclude that this is more likely an adaptive thermoregulation allowing maintenance and slowly rising Tb at a minimal cost, simultaneously with the increasing TA.
In a previous study, the HR in captive black bears was reported to decline gradually over five weeks from the date that food and water were removed [45]. Our finding that changes in Tb began long before changes in HR suggests that studies on captive bears with an artificially defined end of the food/water season might not represent the actual sequence of events in the wild. Our results would have been enhanced considerably had we succeeded at measuring den temperature. Bears in this population are very susceptible to disturbance in winter [46], repeatedly changing dens after captures or capture attempts, so putting temperature loggers inside the den was not realistic.
Our novel results and the methods adapted for this analysis could impact our general understanding of how climate change influences other ecophysiological and behavioral adaptations. In this study, we demonstrate mechanisms for the entry and exit into hibernation by the brown bear in Sweden that have implications for both bear population monitoring and management. These results highlight some of the differences between the bear and small hibernators, reinforcing the importance of not generalizing results from small hibernators to bears.
This work is an example of how different types of datasets can be combined to provide coherent ecophysiological timeseries with potential applications for other ecophysiological and adaptation studies beyond hibernation. Other time-series datasets that could be analyzed in a similar way include phenological and reproduction data on different organisms that are commercially important (crops) or used as indicator species for habitat/ecosystem quality measures (e.g. birds, butterflies). The results from such analyses would provide management strategies and production optimization, while minimizing ecosystem-level impacts. Besides conservation practices, our study demonstrates the importance of several physiological and behavioral characteristics that are important for studies of adaptation, in this case to winter conditions and to climate change, in the context of selection pressures for matching the start and end of hibernation with resource availability.