A newly discovered behavior (‘tail-belting’) among wild rodents in sub zero conditions

Rodents are among the most successful mammals because they have the ability to adapt to a broad range of environmental conditions. Here, we present the first record of a previously unknown thermal adaptation to cold stress that repeatedly occurred in two species of non-commensal rodents (Apodemus flavicollis and Apodemus agrarius). The classic rodent literature implies that rodents prevent heat loss via a broad range of behavioral adaptations including sheltering, sitting on their tails, curling into a ball, or huddling with conspecifics. Here, we have repeatedly observed an undescribed behavior which we refer to as “tail-belting”. This behavior was performed under cold stress, whereby animals lift and curl the tail medially, before resting it on the dorsal, medial rump while feeding or resting. We documented 115 instances of the tail-belting behavior; 38 in Apodemus agrarius, and 77 in Apodemus flavicollis. Thermal imaging data show the tails remained near ambient temperature even when temperatures were below 0 °C. Since the tail-belting occurred only when the temperature dropped below − 6.9 °C (for A. flavicollis) and − 9.5 °C (for A. agrarius), we surmise that frostbite prevention may be the primary reason for this adaptation. It is likely that tail-belting has not previously been documented because free-ranging mice are rarely-recorded in the wild under extreme cold conditions. Given that these animals are so closely-related to laboratory rodents, this knowledge could potentially be relevant to researchers in various disciplines. We conclude by setting several directions for future research in this area.

among the best examples of rodents demonstrating tolerance to a broad range of environmental conditions and thus, comprise populations that are widely-distributed from high to low latitudes of Eurasia. They first ranged from the southern areas of Scandinavia through western, central, and Mediterranean areas to the northern coast of Africa. Later, two ranges were formed; the western range, covering the south-east Scandinavia through the central and eastern Europe to northern Balkans and central Asia; and the far eastern range, from the south of Russia through eastern China, including the Pacific coast. As a result, both species persist in disparate areas across temperate, subtropical, and tropical climates. A. agrarius is also found in the continental climate zones where seasonal and daily temperature fluctuations can range from ca. 30 °C to − 30 °C [17][18][19] .
Given the widespread distribution, these two species could be excellent rodent models for studies on adaptations to extreme temperatures. Such previous studies have included efficient mechanisms of thermoregulation studied at the molecular and sub-cellular levels 20 , basal metabolic rate and thermogenesis [21][22][23] , and behavioral mechanisms such as social thermoregulation 24,25 . In many studies of thermoregulation in endotherms, particular emphasis is given to characteristics of the protruding, exposed parts of the body [26][27][28][29][30] . The presence of exposed organs can be a challenge when the ambient temperature drops below thermoneutrality, thus they must have mechanisms to prevent heat loss. This strategy results in vasoconstriction that reduces blood flow and helps retain heat 31,32 .
Other mechanisms protecting against heat loss are countercurrent heat exchangers closely-spaced vessels, often organized in retes, supplying warm blood to the protruding parts of the body and draining cool blood. This process allows heat to radiate from arterial to venous blood before it reaches the periphery of the protruding organs, where it could be significantly cooled 33 . This approach occurs in such species as sloths 33 , cetaceans 34 and turtles 35 . But among rats (Rattus spp.), countercurrent heat exchangers are not likely to be involved in preventing the loss of heat from tails as it is the case in appendages of some mammals-here a mechanism based on vasodilation/vasocontraction plays a key role 36,37 . Whereas, in Mus musculus, tails appear to contribute little to thermoregulation 38 .
The threats that could result from the destabilization of the body's temperature balance, can also be mitigated behaviorally. In rodents, behavioral adaptations include changes in foraging behavior in the Degu (Octodon degus 39 ) and deer mice (Peromyscus maniculatus 40 ). This seems to imply that, for these species, thermoregulatory abilities may actually be more crucial than mitigating threats from predators 41 . Avoiding thermal stress may also involve modifying essential life tasks, such as finding resources at different times between day and night (desert woodrat, Neotoma lepida 42 ) and different seasons (common vole, Microtus arvalis 43 ). For a recent review of the latter, see 44 . Behavioral thermoregulation is also associated with the exploitation of various thermal refuges, such as vegetation plant cover 45,46 .
Another behavioral phenomenon observed at low temperature is curling into a ball-like posture in order to keep warm, and adopting this posture to reduce the surface-to-volume ratio 47 . Curling, commonly observed in mammals, including domestic pets, has also been described in the rodent literature [48][49][50] . In many animals, such reductions in body surface area also involve setting the protruding parts of the body so that they adhere to its surface as much as possible 51,52 . This behavior is, perhaps, also important to protect them from damage, such as by frostbite, as the trunk temperature is usually higher and kept relatively constant compared to these protruding parts of the body.
A similar phenomenon, which we now refer to as 'tail-belting' , where the animal lifts and may curl the tail medially, before resting it on the dorsal, medial rump, was observed in both A. flavicollis and A. agrarius, during feeding and resting between feeding bouts at our artificial shelter 4 . We collected 4 different temperature 'levels' with A. flavicollis and 5 different temperature levels with A. agrarius. The behavior occurred when the surrounding temperature was below − 6.9 °C. in A. flavicollis, and − 9.5 °C in A. agrarius. However, we only document the appearance of this behavior, and we do not make any apriori hypotheses, nor do we test assumptions. As far as we know, this phenomenon has not been described in the literature, and here we systematically document the occurrence under particular circumstances.

Methods
The observed behavior was recorded during a field study conducted on free-living colonies of yellow-necked mice (Apodemus flavicollis) and striped field mice (Apodemus agrarius) on a private, suburban property in Warsaw, central Poland (52°20′ N 21°03′ E, altitude: 80 m). The experiment took place between 1 November, 2020 and 15 March, 2021 during the Winter season. Temperatures during this period ranged from + 16 °C to − 20 °C. Based on direct and video observations over five months, we estimate the population size of each colony to be in excess of 10 individuals of each species. Individual recognition was sometimes possible based on distinctive variations in coat patterns, body size, and individual characteristics including marks, scars, wounds, variegation of color and shape of tail.
The study included continuous video recording of two chambers which were constructed to test individual responses to scents from conspecifics and/or predators 4  Ethics statement. This observational study was a non-invasive experiment based on the surveillance of free-ranging animals that were free to enter or ignore experimental chambers with food and video cameras. Thus, it did not require permission of the local ethics committee for animal experimentation. The study was carried out on private land with permission of its owners, and all procedures were conducted in accordance with the Polish Animal Protection Act (21 August, 1997). The study was designed and carried out in compliance with the ARRIVE guidelines 56 .

Results
We recorded 115 instances of tail-belting (38 in A. agrarius and 77 in A. flavicollis) during five months of continuous observation of the two colonies. Within the 5-month period, the only instances of tail-belting occurred between January 16 and February 11, 2021 during a particularly harsh winter period in Warsaw, Poland (Supplemental Table 1). Given the number of incidents and the colony size of both species, it is possible that over a dozen animals of each species displayed this behavior. While we could not always identify individuals due to their somewhat uniform appearance, we can be certain that at least 8 individuals (4 of each species) displayed this behavior. We were able to distinguish the 8 individuals due to observable differences in coat patterns, body size, and individual characteristics such as scars, or crooked tails. The lowest chamber temperature recorded during foraging was − 17 °C for A. flavicollis ( Fig. 1; Supplemental Video 1) and − 14.5 °C for A. agrarius (Supplemental Video 2). While animals were recorded across many temperatures during the 5 months, tail-belting was first recorded when the temperature dropped to − 6.9 °C in A. flavicollis and − 9.5 °C in A. agrarius (Fig. 1). Thermal images showed the temperature of the tail dropping well below trunk temperature and, in some cases, equaling ambient temperature below 0 °C for both species (Fig. 2; Supplemental Videos S1, 2). The frequency of tail-belting may have increased with additional decreases in temperature (Supplemental Table 1), though we did not quantify this number.

Discussion
We have documented over one hundred instances of tail-belting, a previously undescribed behavior, among two non-commensal rodents under cold stress in Warsaw, Poland. Because this behavior only occurs in sub-zero temperatures, it is most likely an adaptation to prevent frostbite of an exposed, protruding appendage, or freezing of the tail to a surface. Indeed, we assume frostbite prevention, and not decreased heat loss, is the most likely explanation for two reasons. Firstly, thermovision (Fig. 2, Supplemental Videos 1, 2) showed that the temperature of the tails dropped well below that of the trunk, and equaled ambient temperature even in sub-zero temperatures. Secondly, tails appear to contribute very little to thermoregulation among mice 38 . www.nature.com/scientificreports/ Tail-belting, despite the visual resemblance to common tail curling (the widespread behavior among i.a., cats, dogs, foxes, and lemurs-see 52,57-61 ) seems to be different in function. The purpose of curling tails, especially hairy tails, is to cover other body parts to keep them warm. However, these papers are mainly discussed as observations rather than strictly quantified 52,57-61 . Among larger rodents, a similar tail curl is observed, e.g., in Norway rats and water rats in the cold 62,63 . This behavior is necessary to reduce heat loss by appendages protruding from the body's surface. But here again, there are no data to evaluate the assumptions.
Overall, there is an abundance of information on the contribution of tails to thermoregulation. The tails of rats, beavers, muskrats, foxes, rabbits, and many others contribute to thermoregulation [64][65][66][67][68] . Of course, in these instances, when the ambient temperature drops, the mechanisms guaranteeing the outflow of heat in the tails are turned off, so they cease to function as a radiator. However, there is a clear gap in the data as to what the body does to protect the tails in such conditions, especially when the temperature drops to dangerous values for the tissues that form tails. Apart from behavior data, there is a little information available on antifreeze proteins in mammals' tails 69 .
Given that rodents are among the most successful and best-known animals, particularly the genus Mus 15 , we can only assume this behavior has not previously been documented because of the difficulty of observing small, free-ranging rodent species in sub-optimal conditions in the wild. Additionally, mice minimize foraging in winter while remaining in burrows and consuming hoarded food. Under natural conditions (e.g., without access to our experimental chambers with food), during temperatures when this tail-belting behavior is prominent, most mice would likely not even come out of their burrows. These individuals may have ventured out only because we provided a consistent, aromatic and highly-palatable food on a daily basis for months before and after the cold temperatures of the particularly harsh winter of 2021 in Warsaw, Poland. Indeed, we suspect we only observed this behavior because we set up trials intending to record and assess behaviors in the presence or absence of particular scents near a food reward. Given the relatively distant relatedness between the two species 9,13 it is likely that this behavior also occurs in other rodents, particularly free-ranging animals that usually remain inside burrows in colder climates.
Future studies should be determined by experts in the area of thermal ecology or laboratory researchers interested in this behavior. Examples might include obtaining precise temperature measurements of body parts using thermo-vision cameras 70 as well as morphological and histological comparisons of the species. Precise linear studies should ensue to determine if decreasing temperatures below the threshold increase the frequency of this behavior. Anecdotally, it appears that this was the case. However, we do not know if temperature is the only factor that causes this behavior. It is unknown whether anyone has reduced the temperature in the laboratory in attempts to induce the behavior in other species. M. musculus is currently the primary model for studying frostbite injuries, however, this is not done by lowering the temperature, but instead by adhering frozen magnets to the skin 71 . Thus, tail-belting would not have been expressed by M. musculus under these conditions. Moving forward, this behavior could also be sought out within laboratory conditions, with direct comparisons between the two Apodemus species and Mus.
Mice assays are very popular throughout science and account for more than 60% of all laboratory assays with animals used in research in Europe 72 . Thus, there should be many opportunities to examine whether this behavior is also observed among laboratory animals. We suspect that this behavior is an unconditioned reflex, perhaps if this is shown to be the case, then studies could seek to determine a corresponding neural pathway.