Hibernation (multiday torpor) and daily torpor are used by many endothermic mammals and birds to reduce energy expenditure, often to deal with a shortage of food or adverse environmental conditions (Boyer and Barnes, 1999; Nowack et al., 2017). Torpor in these “heterothermic endotherms” is characterized by a controlled reduction of body temperature (T_b) from high normothermic values, typically ranging from about 36 to 41°C, to values usually between 0 and 29°C and a substantial decrease of metabolic rate (MR) by up to 99% (Boyer and Barnes, 1999; McKechnie and Mzilikazi, 2011; Ruf and Geiser, 2015; Dausmann and Warnecke, 2016).

Torpor expression is strongly affected by size. Torpor is more common in small than in large endotherms and, apparently to maximize energy savings during the torpid state, small species have on average lower minimum T_bs and their MR reductions are more pronounced (Ruf and Geiser, 2015). This seems to be to a large extent related to the fact that, when normothermic, small species have to compensate for high heat loss via a relative large surface area, have a limited capacity for fat storage and high costs for locomotion (Tucker, 1975; Calder, 1996). Moreover, because the lower critical temperature of the thermo-neutral zone (TNZ) of small species is often around 25 to 30°C and ambient temperature (T_a) experienced in the wild is generally below that, they typically have to produce heat endogenously because of the large T_b–T_a differential (Tattersall et al., 2012; Kronfeld-Schor and Dayan, 2013; Riek and Geiser, 2013; Withers et al., 2016; Mitchell et al., 2018).

Although almost all mammals and birds are competent endotherms as adults, the vast majority are altricial at birth or hatching. Altricial neonates or hatchlings are small, naked and uncoordinated and, when removed from the nest or away from their parents, cool rapidly to near ambient temperature (T_a) because they are only partially endothermic and unable to produce sufficient endogenous heat for maintenance of a high and constant T_b (Dawson and Evans, 1960; Morrison and Petajan, 1962; Holloway and Geiser, 2000); therefore they are often referred to as being poikilothermic (Greek “poikilo” = variable, “therme” = heat). As these animals grow to larger size and fur or feathers develop together with a maturation of the nervous and other organ systems, the ability for endogenous heat production and heat retention improves and eventually they are able to maintain a constant high T_b, at least during exposure to moderate T_as. These newly endothermic young, typically at about 20–50% of adult body mass, are highly vulnerable to heat and energy loss or even death by starvation when they maintain a high constant T_b at times the parents are unable to supply enough food and because of their inexperience in obtaining food on their own (Geiser, 2008).

Potentially, using torpor during periods of high heat loss and insufficient food supply would provide an avenue to enhance survival of young. However, although such energetic bottlenecks are likely faced occasionally or even frequently by the majority of small young mammals and bird species, as for example during bad weather, information on torpor during development is limited to only a few species (<15 species or <0.1% of the >15,000 mammals and bird species), including some small marsupial mammals, a few placental mammals, and a few birds (Nagel, 1977; Geiser, 2008; Eichhorn et al., 2011; Giroud et al., 2014; Wacker et al., 2017). For most of the studied species the rate of development is rather slow, and, especially in marsupials, for which most detailed data are available, first expression of torpor is observed only months after birth following a prolonged poikilothermic phase (Geiser et al., 2006, 2017; Wacker et al., 2017). The slow development in many species expressing torpor is supporting the view that heterothermic species often have a slow life history (Turbill et al., 2011), typically producing only 1 or 2 litters/young per year.

The purpose of our investigation was to provide the first data on the development of endothermy and torpor expression in small desert hamsters (Phodopus roborovskii; adult BM ∼25 g). This species is interesting with regard to its reproductive and developmental biology as it differs from most other species in which development of torpor has been investigated. It produces up to four litters of up to ten young per year, can breed within 2 months of birth, lives for only 1 or 2 years in the wild, and is clearly r-selected (Kolynchuk, 2015). Desert hamsters are distributed over the harsh deserts of continental Asia and, as adults, use daily torpor to aid survival in their environment (Chi et al., 2016). As they reproduce well into autumn, their young are potentially exposed to adverse environmental conditions and energy shortages in late autumn/early winter, and may require the use of energy savings mechanisms such as torpor to survive over winter into adulthood. We therefore tested the hypothesis that this fast-growing species is able to express torpor early during the growth phase, soon after endothermy is attained. Such data will improve the understanding of the functional significance of torpor especially in rapidly developing altricial mammals and the role of heterothermy during development in endotherms in general.

The rate of development of desert hamsters is astonishing. Endothermy is reached as early as at 15 days of age at a body mass of only 5.5 g and already at 16 days (BM 5.8 g) the animals are able to express torpor with a rapid decline of MR at torpor entry and the ability of rewarming from low T_b at the end of the torpor bout.

The development of thermoregulation and morphology was qualitatively similar to what has been observed in other altricial birds and mammals (Morrison and Petajan, 1962; Geiser, 2008). Initially when the animals were small and naked, rates of cooling were high. Rates of cooling decreased with an increase in size, improvement of insulation, but likely also an improvement of non-shivering thermogenesis from brown adipose tissue (BAT) (Oelkrug et al., 2015). The difference between our and most other studies was the speed of development with competent endothermy in desert hamsters reached only after 15 days, much faster than in many other species, especially small marsupials that take several weeks to reach that state (Geiser et al., 2006; Wacker et al., 2017). However, the values were similar that what has been observed in the related and similar-sized Djungarian hamster (Phodopus sungorus) (Bae et al., 2003).

The fast development of thermoregulation of desert hamsters was followed by the ability to enter and arouse from torpor at an age as young as 16 days (BM 5.8 g). Although we quantified torpor at a rather mild T_a because of consideration of the animals’ wellbeing, lower T_as likely would have resulted in increased torpor use (see Chi et al., 2016) especially later during development. Torpor was characterized by a reduction of MR to <10% of the MR during activity, which was as high as that of bats and birds during flight, which are among the highest MR measured for vertebrates (Speakman and Racey, 1991). Rewarming from low T_b during torpor was achieved by juvenile desert hamsters although heat production is reduced at low T_b and although the high relative surface area at a small BM facilitates heat loss and makes endogenous rewarming difficult (Currie et al., 2015; Wacker et al., 2017). The fast development of torpor differs again from small marsupials, which reach that state only after several weeks (Wacker et al., 2017), but also from the related P. sungorus, which expressed fasting-induced torpor only from the age of 28 days (Bae et al., 2003). This difference may appear surprising considering the similar size, similar development of endothermy and similar general biology of the two congeners. However, the discrepancy is likely explained by the commencement of measurements of fasting-induced torpor in P. sungorus, which did not begin until day 21, soon after the animals had been implanted with transmitters. Therefore, it is possible that, if P. sungorus had been measured earlier, they would be rather similar to desert hamsters, an assumption that is supported by the observation of 2-deoxy-D-glucose (2-DG) induced T_b reduction in P. sungorus at the age of 16 days (Bae et al., 2003). Although it has been argued that the 2-DG induced reduction in T_b differs from natural torpor (Westman and Geiser, 2004, but see Chi et al., 2018) it nevertheless shows that the animals were able to rewarm from low T_b at a very early age.

What is the likely function of torpor during development? Constant endothermy in small species is extremely costly because of the high heat loss and requires a more or less uninterrupted supply of energy in the form of food. If adults are unable to supply the food or juveniles unable to obtain it, they are in danger of starvation or even death. This can be overcome by the expression of torpor, which reduces energy and therefore foraging requirements substantially (Nowack et al., 2017), and can prolong the time over which stored energy in the body can be extended, in extreme cases for up to several months (Ruf and Geiser, 2015). Of course huddling in nests also will reduce energy expenditure to some extent and is important during the development (Gilbert et al., 2010). However, it does not prevent occurrence of hypothermia in groups of young in the wild (Holloway and Geiser, 2000). Moreover, the limitation of huddling lies with its inability of reducing MR below basal levels, whereas during torpor MR values well below BMR can be reached and energy loss can be minimized (Namekata and Geiser, 2009). While survival is clearly an important function of torpor during development it also has the potential to be used for diverting nutrients to facilitate growth as growth rate on limited food is enhanced by torpor use (Giroud et al., 2014). Since the vast majority of mammals (for example, marsupials, bats and many rodents) and birds (for example passerines) are altricial at birth/hatching and will have to undergo similar morphological and functional changes described here, the expression of torpor during the development may be far more often employed that is currently appreciated, and may even include some small precocial species.

Past data suggested a different development sequence in torpor expression between birds and marsupial mammals on one hand and placental mammals on the other (Geiser, 2008). In birds and marsupials torpor expression seemed to develop immediately after competent endothermy was attained (i.e., heterothermic endothermy develops immediately after poikilothermy). In contrast, as outlined above, the data on the placental rodents suggested that the initial partial endothermic phase (poikilothermy) is followed by an intermediate homeothermic phase with constant high T_b lasting for about 2 weeks in hamsters (P. sungorus) with the first bout of torpor after food restriction observed only at the age of 28 days (Bae et al., 2003). More extreme, in juvenile ground squirrels (Spermophilus saturatus) this homeothermic phase is prolonged and torpor is expressed only ∼4 months after birth (Geiser and Kenagy, 1990). However, data from our present study with torpor expression in placental desert hamsters only 1 day after reaching endothermy suggest that there may be no generic difference between marsupial and placental mammals and that perhaps the pattern observed in ground squirrels is a reflection of their extreme seasonal thermal biology characterized by homeothermy in spring/summer and heterothermy only from late summer through autumn/winter. Thus, while our data do not support the view that there may be common developmental differences with regard to torpor expression among endothermic classes or sub-classes they nevertheless suggest that it is widely used by many small endotherms.

It is interesting that torpor expression is reduced with growth. Likely this is related to the reduction of mass-specific RMR in thermo-neutrality (Figure 5), which interestingly, despite the young were growing and should have a raised RMR, had a slope of –0.33 as that measured for BMR in fully grown adult mammals (White and Kearney, 2013). However, and perhaps more importantly with regard to our study, the reduced torpor expression with growth was likely caused by the exposure to only a mild cold load (T_a 22°C), a reduction on RMR during cold exposure, and a reduced cost of locomotion that decreases with size (Tucker, 1975). Especially the mild T_a differs substantially from those experienced by the animals in the wild in winter when they are exposed to very low T_a and food is probably scarce. Therefore individuals in the wild likely will express torpor even when they reach adulthood (Chi et al., 2016).

Clearly heterothermy during development provides juveniles with flexible thermo-energetics and consequently a better chance of survival, as is also the case even before birth during reproduction (Willis et al., 2006). Considering how few species have been studied in this regard, it is possible that torpor during development is used by many if not the majority of altricial species. This interpretation is supported by the fact that shallow torpor during development has been observed even in species considered to be homeothermic as adults such as in juvenile storm petrels (Oceanodroma furcata) and laboratory rats (Rattus norvegicus) (Boersma, 1986; Nuesslein-Hildesheim et al., 1995). It therefore seems an omission that the phenomenon of torpor use during development has not been more widely examined with the aim to improve the understanding of its diversity and function.

This study was carried out in accordance with the recommendations of Animal Care and Use Committee of Institute of Zoology, the Chinese Academy of Sciences. The protocol was approved by the Animal Care and Use Committee of Institute of Zoology, the Chinese Academy of Sciences.

FG finished the main parts of the manuscript. Q-SC, JW, GS, and FG conducted the experiments. FG, Q-SC, and D-HW conceived this study.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.