THE EFFECT OF HUDDLING ON THERMOREGULATION AND OXYGEN CONSUMPTION FOR

THE NAKED MOLE-RAT

P. C. WITHERS* and J. U. M. JARVB Department of Zoology, University of Cape Town, Rondebosch, South Africa 7700

(Receicd I4 Auqust 1979)

Abstract-l. Individual and huddled naked mole-rats became progressively hypothermic at lowered ambient temperatures; & = 20.6 + 0.41 T..

2. Oxygen consumption of individual mole-rats increased more than that of huddled mole-rats at T, < 34’C.

3. Huddling behaviour considerably reduced the mass-specific thermal conductance from 0.25 ml

O2 g-’ ‘C-’ (individual) to 0.12 (group of four) at T, = 2O’C. 4. Thus, huddled mole-rats are unlike other endotherms since they maintain the same & as individuals

but at a reduced metabolic cost. 5. Huddled mole-rats had lower rates of evaporative water loss than individuals, at high T,, because of

their lower exposed surFace area and Vo, (per individual).

INTRODUCTION

The naked mole-rat (Hetrrocephdus glaber Ruppell) is effectively naked, having only tactile vibrissae, and has a poor ability to thermoregulate perhaps because of the lack of insulation (Thigpen, 1940; Hill et al., 1955 ; McNab, 1966). The naked mole-rat is probably the most ineffectual (adult) endotherm living today, even in comparison with monotremes, marsupials and primitive eutherians. Heterocephalus is not, however, a primitive mammal in an evolutionary sense since it is a highly-specialized bathyergid rodent whose living relatives are endotherms, and whose ancestors would have been endotherms. The poor thermoregulatory capabilities of Hetrrocrphalus therefore appear to be a recent development, in response to its thermostable environment (McNab, 1966).

Hrtrrocrphalus is also a highly-specialized rodent in other respects. Naked mole-rats are extremely colo- nial. with often more than 40 individuals in a single colony, they build permanent burrow systems with extensive tunnels and communal nest chambers, and they have a complex social hierarchy (Jarvis & Sale, 1971; Jarvis, 1978). We have therefore reinvestigated the thermoregulation and energetics of naked mole- rats in view of these communal habits. We report here aspects of body temperature regulation, rates of oxy- gen consumption, and rates of total water loss for individual mole-rats, for mole-rats in groups of two and four, and for mole-rats with artificial insulation.

METHODS

The naked mole-rats investigated here were captured at Mtito Andei, Kenya, and maintained in the laboratory at Cape Town at temperatures and humidities similar to those of their natural environment Experiments were con- ducted with the animal, or animals, in clear plastic meta- bolic chambers (volume = 3000cm3) in a constant-

* Present address: Department of Biology, Portland State University, Box 751, Portland, Oregon 97207, U.S.A.

remperature chamber which allowed maintenance of ambient temperatures to within f 1°C. Ambient air tem- peratures within the metabolic chambers were measured with a thermocouple pushed into the chamber through the air excurrent port. Experiments were conducted between 15 and 34°C.

An open-flow respirometry system was used for all oxy- gen consumption and water loss experiments. Air flow rate (ATP) was measured with G.E.C. Elliott Rotameter flow- meters which were calibrated prior to use. The oxygen, carbon dioxide, and water contents of the excurrent air were determined using a Beckman OM I4 oxygen analyser in the following manner. Air samples were removed from the excurrent air stream using 60cm3 syringes, and then injected through the oxygen analyser with the CO, and H,O removed (using ascarite and drierite). with only the Hz0 removed, and with CO* and H,O present (see Withers, 1977). Rates of oxygen consumption (r’o,; ml O2 g-’ hr-‘), respiratory quotient, and rates of total water loss (TWL; mg H,O g- ’ hr- I) were calculated using stan- dard equations (Withers, 1977). All values are STPD.

Body temperatures (Tb; “C) of mole-rats were measured with fine thermocouples sheathed in PE tubing, using a Bailey Batt 4 thermocouple meter. Mole-rats were not post-absorbtive for the metabolic measurements because withholding food made them restless.

Burrow microclimate was investigated at Mtito Andei. All temperature measurements were made with Bailey Batt and YSI thermistor meters. Relative humidity (rh; “/,) of air aspirated from burrows through a length of PE tubing was measured with a combination wet-dry thermocouples.

All statistical values are presented as mean f standard error, with the number of observations.

RESULTS

The naked mole-rats tended to be very restless, par- ticularly when only one animal was in the metabolic chamber, and considerable patience and care were required to obtain data from resting animals. Further- more, mole-rats did not invariably huddle together when given the opportunity to do so, and this con- tributes to the scatter of data for animals in groups.

215

216 P. C. WITHERS and J. LJ. M. JARVIS

Table 1. Body temperature gradient (Th - 7& rate of oxygen consumption (I&,: ml 0, g- ’ hr-‘), eva- porative water loss (EWL; mgH20g-’ hr-‘) and “wet” and “dry” thermal conductance (C,, C,; ml Oz g- ’ hr- ’ ‘Cm ‘) for individ~~1 and grouped naked mole-rats at ambient temperatures from 15 to

34°C. Values are % k SE (II) or Ti. (it)

(r, - r,) r’oi EWL c; C,T

N = 1 1.15 4 0.07(4) 0.62 f 0.04(15) 7.4 + 0.6 (I 5) : : 34°C N=2 -0.15 * 0.35 (4) 0.67 f 0.06 (8) 5.4 + 0.8 (8) :: ::

iv = 4 -0.06 k 0.15 (8) 0.45 ) 0.02 (I I ) 2.X i 0.6(11) : :

/L’ = 1 3.5 k 0.20 (8) 1.42 f 0.06 (39) 3.3 & 0.2 (39) 0.39 + 0.02 (8) 0.26 i: 0.03 (8) 3o’c it’ = 2 3.4 i: 0.24 (8) 0.85 i 0.05 (21) 2.4 + 0.3 171) 0.25 * 0.03 (4) 0.18 2 0.02 (4)

M = 4 3.0 * 0.1714) 0.79 k a.07 (2 I ) I.3 * 0.1 (70) 0.23 (2) 0.19 (2)

K = 1 6.9 (2) 2.52 ~O.l3(li) 3.4 i 0.6(1 I) 0.38 (2) 0.31 (2) 25°C N=2 5.4 rf: 0.20 (3) 1.70 t 0.14(24) 2.5 + 0.2 (24) 0.24 i 0.09 (3) 0.20 + 0.07 (3)

N = 4 4.5 i: 0.39 (3) 1.45 f 0.15(16) 1.7 * 0.2(12) 0.27 f 0.07 (3) 0.20 rt: 0.07 (3)

N=l 9.7 & 0.20(4) 2.65 + 0.08 (22) 1.5 i_ 0.2 (8) 0.27 + 0.01 (4) 0.25 (2) 20°C N = 2 9.7 & 0.28 (8) 1.52 + 0.15 (20) 1.8 f 0.3(11) 0.20 & 0.03 (4) 0.14 (2)

N = 4 9.6 & 0.14(13) 1.26 + 0.12(32) 1.6 i 0.2(10) 0.12 + 0.03 (4) 0.12 (2)

IV = 1 13.2 & 0.92 (3) 2.74 f 0.08 (14) 0.21 * O.Ol(3)

iS’C N = 2 12.1 & oso(6r 2.29 & 0.32 (10) 0.20 (1) N = 4 9.8 + 0.45(12) 0.60 i: 0.04 (14) 0.06 * O.Ol(3)

* C, = ri,,/(r, - T,). t Cd = C,. - [EWL/(7, - K)] assuming I ml O2 = 4.8cal and latent heat of evaporation is

580calg~ ‘. : Tb z T,.

The G of individual mole-rats varied from about 26’C at an ambient temperature (T,) of IYC, to 34°C at z = 34°C (Table I). The Tb of individual and grouped mole-rats varied with T, as follows:

r, = 20.q * 0.4) + 0.4 I( * 0.02) T,

(r’ = 0.80, A’ = 171)

There were no differences in & of mole-rats with or without artificial insuiation and there were no differ- ences in 5 of individual and grouped mole-rats at 7; = 20 and 3o”C, and slight differences at ‘&, = 15, 25 and 34°C. The 7$ which we report for naked mole-rats at high T, are similar to those reported by McNab (1966) but our values are much higher than McNab’s at T, = 15 and 2O’C (Fig. I).

Amblent temperature (“C I

Fig. I. Body temperature of naked mole-rats at ambient temperatures from 1%34°C. Values are mean +SE with the ranges also indicated. Broken line is body temperatures from McNab (1966). Solid block indicates values for mole-

Fig. 2. Rates of oxygen consumption for mole-rats at ambient temperatures from 1$34iC, when alone (I) and in groups of two (2) and four (4). The maximum rates of oxygen consumption for individual mole-rats (l?ob,,,.) and minimum values for quiet, closely huddled groups of four mole-rats ( l&,,,i,] are also shown. Circles indicate the rates of oxygen consumption for individual mole-rats (I) con- verted to a constant body temperature of 34°C (the body temperature at an ambient temperature of 34~C) assuming a Qio of 2. Values are mean &SE (SE for (2) is omitted for

rats under natural conditions. clarity).

Amblent temperature, OC

Thermoregulation of naked mole-rats 217

The f&, for individual mole-rats increased at T, < 34°C but the rate of change in I&, with T, was less at T, < 25°C (Fig. 2). The minimum i$o, for indi- vidual mole-rats at T, = 34°C (presumably a thermo- neutral temperature; McNab, 1966) was 0.62 + 0.04 (N = 15) which is similar to the value of 0.55 reported by McNab (1966). However, our data for liol of mole-rats at T, < 34°C are generally less than those reported by McNab (1966) despite our mole- rats having much higher T,s. The maximum f&, values determined for individual mole-rats are also included in Fig. 2.

The voio, for mole-rats in groups of four and two was markedly less than that for individual mole-rats, even though Tbs were similar (Fig. 2). The liol for groups of four mole-rats increased at T, = 30 and 25°C relative to that at 34°C. but declined at lower T,s. However, the lowest values of I&, for groups of four mole-rats (which were closely huddled) was inde- pendent of &, at 0.3-0.4 ml O2 g- ’ hr ’ (Fig. 2).

The effect of providing cotton wool as artificial nesting material was investigated at T, = 25 and 30°C. Individual mole-rats with cotton wool had lower Ijoz values than individuals without cotton wool (but the same 7&z), and groups of two and four mole- rats had decreased I&, if cotton wool was provided (Table 2).

Physical laws for heat exchange (i.e. Newton’s Law of Cooling) state that the temperature of a body at equilibrium is determined by the rate of heat exchange, the difference between body and ambient temperature. and the thermal conductance i.e. for an animal, HP = C(T, - TJ where HP is the rate of heat production (or I& with an appropriate conversion factor), C is the thermal conductance. The value of C for naked mole-rats can be measured in a number of ways. The value of C for a dead mole-rat resting on a poor conductor (3 in. of cotton wool) or suspended in air, was determined from cooling curves to be 0.38 and 0.74 ml O2 g- ’ hr- ’ [“C]- ’ respectively. C could also be estimated from Fig. 2 for live individuals as the slope of the line for PO, if their body temperature was constant. Such values for C are 0.2-0.3 ml 0, gg ’ hr-’ [“Cl-’ if the f&:b, values for individual mole-rats are recalculated to a Tb of 34°C assuming a

Q 1o of 2 (Gesser et al., 1977; Fig. 2). Alternatively, C can be calculated as i&,,/(T, - T,) and such values for individual mole-rats are from 0.2 to

0.4 ml O2 g ’ hr- ’ [“Cl- ‘. The predicted C for a mammal with the same body mass as a naked mole- rat (40 g) is 0.16 ml 0, g- ’ hr- ’ [“Cl- r (Morrison et al., 1959).

The preceding values for C include heat lost through evaporation, and thus are “wet” conduc- tances (C,). Values for “dry” conductance (C,) can be obtained by subtracting evaporative heat loss via evaporation from metabolic heat production. Values of C, and C, for naked mole-rats under the various experimental conditions are summarised in Table I and Table 2.

The TWL of individual mole-rats increased drama- tically at the higher T,s (Table l), and the TWL potentially dissipated 1300/;, of the metabolic heat pro- duction at G = 34°C. The TWL for mole-rats in groups of two and four were similar to (at T, = ZO’C) or greater than (at T, 25°C) the TWL of individuals (Table I). The TWL of groups of four mole-rats at Th = 34’C potentially dissipated 70% of the metabolic heat production.

The mean T, for 45 mole-rats immediately after capture at Mtito Andei was 33.9”C (range: 3 1.7-36.2”C). Burrow temperatures varied slightly, depending upon depth and time of day, but were rela- tively constant at about 32’C (range: 3&34”C). There was no daily variation in soil temperature at a depth of 100 cm (T = 31.3”C) whereas soil temperature varied by 0.9”C (3 1 .I-32°C) at SO cm and by 6.4”C (30.4-36.6nC) at 5 cm. S,ufzace soil temperature varied markedly during the day from 32 to 63°C whereas air temperature was from I6 to 33°C. Relative humidity within burrows was 92;; & 1.2 (;I; = I I) with a range of 85--loo”,,.

DISCUSSlOli

Heterocepfdus are unique among small (adult) mammals in being effectively naked. This lack of insu- lation potentially has serious thermoregulatory conse- quences for such a small mammal (40g) which has a high surface:volume ratio, and might also affect water balance if a naked skin evaporated more water than a skin with fur.

Individual mole-rats are clearly poor thermoregu- lators in comparison with other mammals, with & changing in a linear fashion with T,. A similar lability of T, was, surprisingly. found for mole-rats huddled in

Table 2. Comparison of rate of oxygen consumption ( Ijo2; ml O2 g _ ’ hr _ ‘) and dry thermal conduc- tance (Cd; ml O2 g- ’ hr- ’ Y-‘) for individual and grouped naked mole-rats with. and without, artificial

nesting material at ambient temperatures of 25 and 3O‘C. Values are mean &SE (N)

No nesting material With nesting material v o2 Cd* v 02 C*i

30°C

25°C

N=l I .42 + 0.06 (39)** 0.26 k 0.03 (8) 0.83 + 0.08 (28) 0.12 k 0.02 (6) N = 2 0.85 + 0.05 (21)** 0.18 & 0.02 (4) 0.55 + 0.05 (I 2) 0.09 (3) N=4 0.79 f 0.07 (2 1 )“s 0. i 9 (2) 0.59 * 0.05 (1 if 0. I 2 (2)

N=l 2.52 & 0.13 (1 l)* 0.31 (2) 2.09 & 0.13(7) 0.23 (2) N=2 1.70 + 0.14 (24)Ns 0.20 + 0.07 (3) 2.00 -t: 0.20(21) 0.18 (3) N=4 1.45 & 0.15 (16)** 0.20 + 0.07 (3) 0.45 + 0.05 (11) 0.06 (2)

t Calculated as indicated in Table 1. Differences between means for oxygen consumption with, and without, nesting material are significant at P < 0.05 (*). P < 0.001 (**) or not significantly different (NS) using the t-statistic.

218 P. C. WITHERS and J. U. M. JARVIS

groups of two and four. For all mole-rats, & = 20.6 + 0.41 T,. This thermolability confers an energy savings of up to 50y0 (,of the required meta- bolic heat production to maintain constant ‘&) at low T,. albeit at the cost of thermal homeostasis.

The relationship between voo2 and T, for individual mole-rats was qualitatively similar to that seen for other mammals, but the extent of the increase in metabolic heat production was insufficient to main- tam T, constant. The change in poz per “C decrease in T, was less at K < 25’C than at T, > 25°C reflecting the Ilypothermia at low T,s. If the values for I;b, of individual naked mole-rats at T, 34°C are corrected to a & of 34°C using a Qlo of 2 (see Gesser ef al., 1973, then the thermogenic response of naked mole-rats is more like that of other mammals (Fig. 2).

The maximum li, values for individual mole-rats were about 5ml O2 g-’ hr-’ at T, = 25°C (Fig. 2). This is less than the predicted i&2m.i for a 40g mam- mal of 6.1 ml O2 g- ’ hr- ’ (Lechner, 1978). but the mole-rats only had a & of about 32’C. If the J$, of active mole-rats is converted to a Th of 37”C, this value of 7 ml 0, g- ’ hr -- ’ exceeds the predicted value of 6. This indicates that naked mole-rats have a typical mammalian metabolic capacity in uivo, as well as irt vitro (Gesser er al., 1977).

Huddling behaviour has been shown to reduce the metabolic cost of thermoregulation. at constant T,, for mammals and birds (Pearson, 1960; Baudinette, 1972; Glasser & Lustick, 1975; Pinshow ef ul., 1976) or to result in higher Th for clustered. torpid bats (McNab, 1974; Howell, 1976). A thermoregulatory advantage of huddling is not limited to birds and mammals, but is also apparent for reptiles (Lasiewski & White, 1971) and insects (Nagy & Stallone, 1976). Huddling naked mole-rats, like other mammals, maintain similar Tbs as individual mole-rats, but at a reduced metabolic cost. However. naked mole-rats, in marked contrast to other mammals, do not maintain the same, high ?&, (eg. 34’C) but become hypothermic whether indivi- dual or huddled, at low T,. The i’oz of cold-stressed, huddled mole-rats is lower than that of individuals at any particular T2, and the minimum J$ol of huddled mole-rats was actually independent of To. These data suggest that naked mole-rats maintain a certain Tb which is determined by the T,, and that huddling is used to gain an energetic saving, rather than a thermoregu~atory saving. This is a unique “strategy” amongst mammals, but it is not possible to interpret these data in an ecological context at present since nothing is known concerning the energetics of free- living naked mole-rats.

The similarity of I&, for individual and huddled mole-rats, but the marked differences in I&,. indicate disparate values for thermal conductance (C). The value of C for a dead mole-rat is greater than that of live animals, perhaps reflecting the importance of blood-flow patterns (Table 1). The values of C for individual mole-rats, calculated as &/(T, - x) varied at different T,s, indicating a marked departure from Newton’s Law of Cooling. The decrease in lie, (corrected to a Tb of 34°C) at low T, also indicates a decreased C. This decrease in C most likely is due to peripheral hypothermia. If an animal allows its body extremities to cool below core Th (regional hetero- thermy}, the value of C is lowered because G (core)

and T, are the same for animals with or without regional hypothermia, but the Po, of the indi~~idual with hypothermy is less (since part of its body is col- der than core 7i). Hence, C = ~o,/(Tb - &) is less. The lowered values of C for individual mole-rats (from 0.41 ml O2 g-’ hr -I [“Cl-’ at 7;; = WC to 0.21 at T,, = 15OC) indicate extensive peripheral hypothermy. Naked mole-rats do not have a layer of subcutaneous fat but have fat bodies in the abdominal cavity (Thigpen, 1940). Naked moie-rats do not. there- fore, tolerate regional hypothermia within an insula- tive layer of fat, as marine mammals do (Scholander rr al., 1950).

There were striking differences in values of C for individual mole-rats compared to huddled animals, and animals with artificial insulation (cotton wool). The predicted C for four naked mole-rats huddling effectively (i.e. body mass = 4 x 40 g) is 0.08 ml O2 g- ’ hr- ’ [“Cl - ’ (Morrison et al., I95 I). The cal- culated C (as @o’/(T, - T,) for four huddled mole-rats varied from 0.27 at T, = 25°C to 0.06 at T, = IYC, indicating effective huddling and perhaps also some regional hypothermia at low T,s.

The rates of water loss (TWL) for naked mole-rats varied from 0.6mg Hz0 [ml O,] _’ at T, = 20°C to 12 at T, = 34°C for individuals, and from 1.2 to 6 for mole-rats in groups of four. These values are con- siderably greater than the rates of pulmocutaneous water loss for other rodents (MacMilien, 1972). How- ever, because of the experimental design in the present study (sometimes providing cotton wool), it was not possible to exclude urination as one component of water loss. Nevertheless, the TWL for individuals was similar to (at low T,) or less than (at higher T,) that of mole-rats in groups of four, as might be expected. Both the lower lioL and lower exposed surface area (per individual) of the huddled mole-rats would de- crease TWL (at least the pulmo-cutaneous com- ponent). The ambient relative humidity would have been similar for the individual and huddled mole-rats (since we adjusted the air flow-rate to give a change in per cent of O2 of about 0.5 for individual and huddled mole-rats), so we believe that the lower TWL of hud- dled animals is a result of huddling per SP. Baudinette (I 972) also reports lowered evaporative water loss for huddling rodents.

Huddling is unlikely to be significant to the water balance of naked mole-rats (despite their nakedness and probably higher rates of water loss) because of the high burrow humidities in the field. The rh in a natural burrow, if not already saturated. would rapidly become saturated in the vicinity of mole-rats because of their high TWL. However, a group of naked mole-rats should saturate their immediate en- vironment quicker than an individual.

It is apparent from this study, and that of McNab (1966), that naked mole-rats live in an environment (at Mtito Andei) which is thermally stable and humid, and so are not normally exposed to hot, cold or desic- cating conditions. The present study indicates, how- ever, that naked mole-rats can nevertheless regulate their Tb to a greater extent through metabolic heat production, evaporative water loss, huddling behav- iour, and use of artificial insulation, than was pre- viously indicated by McNab (1966). Naked mole-rats in nature are probably homeotherms, even if they are

Thermoregulation of naked mole-rats 219

not effective endotherms. It is therefore not surprising that the temperature sensitivity of enzymes from naked mole-rats, and in vitro tissue metabolism, are similar to those of other mammals (Gesser et al., 197X).

It is of more than simple academic interest to study the thermoregulatory capabilities of ~eter~cep~u~~~ over a wider range of T, than are normally experi- enced at Mtito Andei. An uncommonly cold season, for example, could be of great adaptive significance to naked mole-rats. Populations of naked mole-rats at high elevations (eg. Isioio; Hill pf al., 1957) or at the periphery of their geographic range would be more influenced by ambient temperature than those at Mtito Andei. Furthermore, Hrtrrocvphalus can pro- vide unique insights into the potential thermoregulat- ory capabilities of primitive, small mammals even though Ifeterocephalus is an advanced. specialized bathyergid rodent whose ancestors were undoubtedly endotherms.

~~~~ro~e~~z~~~s is uniquely different to the extant, so-called “primitive” mammals in that it has the nor- mal metabolic capacity of a mammal, but virtually lacks insulation. The “primitive” mammals such as monotremes, marsupials, and golden-moles have nor- mal mammalian pelage (and hence normal values for thermal conductance) but tend to have low metabolic rates (see Kinnear & Brown, 1975; Schmidt-Nielsen et al., 1966; Grant & Dawson, 1978; Withers, 1978). These “primitive” mammals are so-called by virtue of their metabolic machinery, not their insulation.

The evolution of endothermy is generally con- sidered in terms of the evolution of insulation, and the thermal effects of large body mass (high thermal iner- tia and passive homiothermy) (see McNab, 1978). However, a significant difference between mammals and ectotherms (such as lizards) is the elevated meta- bolic capacity of mammals (indicated by higher rest- ing and maximal rates of oxygen consumption). This quantum change in metabolic heat production capa- bility of mammals was perhaps the primary basis for the evolution of endothermy by an animal which was not necessarily very large. Whether or not this is so, the naked mole-rat is the only extant example of a small mammal with a normal mammalian metabolic capacity, but a non-mammalian thermal conductance (C, for an individual naked mole-rat at T, = 30°C is similar to that for a 40g lizard). The present study clearly demonstrates that the selection of a suitable thermal-environment, use of artificial insulation, huddling behaviour, peripheral hypothermia and a mammalian level of tissue metaboIism, enable naked mole-rats to be endothermic, and partially homio- thermic, even without an insulative layer of fur.

Ac~nowlrdyemrtrrs-We gratefully acknowledge the sup- port and loan of equipment given by Professors G. N. Louw and W. R. Siegfried. We thank Subu and Ndalinga for energetic assistance in the field, and Dr and Mrs Baliy for their hospitality. One of us (PCW) gratefully acknowl- edges a University of Cape Town bursary and research grant. We thank J. Jones for critical comments of the manuscript.

REFERENCES

BALIDINETTE R. V. (1972) The impact of social aggregation

on the respiratory physiology of Australian hopping mice. fom~. &o&em. Phvsial. 41A. 3-%3X.

GESSER H.. jOHAlriSSEN K.‘& MALOIY G. M. 0. (1977) Tissue metabolism and enzyme activities in the rodent Heterocephalus glaber, a poor temperature regulator. Camp. B&hem. Physiol. 57B, 293-296.

GLASSER H. & LUSTICK S. (f975) Energeetics and nesting behavior of the northern white-footed mouse, Pero- m~scus 1eucopu.s rioveboracensis. Pilysiol. Zool. 48, 105-I 13.

GRANT T. R. & DAWS~N R. J. (197X) Temperature regula- tion in the platypus, ~~~1it~z~~~I~ft~c~s arratinus: production and loss of metabolic heat in air and water. P~~~~sio~. Zool. 51, 315-332.

HILL W. C. O., PORTER A., BLOOM R. T., SEAC~ J. & SO~JTHWICK M. D. (1957). Field and laboratory studies on the naked mole-rat, ~~~?~~oc~p~~~~~u,~ yinher. Proc. 3001. sot. Lorrd. 128, 455.-5 13.

HOWELL D. J. (1976) Weight loss and temperature regula- tion in clustered versus individual GIossopllaga soricina. Camp. Biochem. Physiol. 53A, 197- 199.

JARVIS J. U. M. (1978) The energetics of survival in Heturo- eepbu~~~s glaber (Ruppeil), the naked mole-rat (Rodentia: Bathygeridae). Canlrgie Mus. Nut. Hist. Bull. 6, X1-87.

JARVIS J. U. M. & SALE J. B. (1971) Burrowing and burrow patterns of East African mole-rats Tachyorycres, Helio- pbobius and ~efer~c~~~~u~l~s. J. Zooi.. Lo&. 163, 451-479.

KINNEAR A. & SHIELD J. W. (1975) Metabolism and tem- perature regulation in marsupials. Camp. Biockem. Phy- siol. 52A. 235-245.

LE~HNER A. (197X) The scaling of maximal oxygen con- sumption and pulmonary dimensions in small mammals. Respir. Physiol. 34, 29-44.

MACMILLEN R. E. (1972) Water economy of nocturnal desert rodents. Symp. mol. sot. Land. 31, 147-I 74.

MCNAB 8. K. (1966) The metabolism of fossoriat rodents: a study of convergence. ecology 47, 712-733.

MCNAB B. K. (1974) The behavior of temperate cave bats in a subtropical environment. Ecolog, 55, 943-958.

MCNAB B. K. (197X) The evolution of endothermy in the phylogeny of mammals. AHI. Nut. 112, I-21.

MORRI~~ P. R.. RISER F. A. & DAWE A. R. i I9591 Studies on the physiology of the masked shrew Sorrs ciwreus. Physiol. Zool. 32, 256-27 I.

NAGY K. A. & STALLONE J. N. (1976) Temperature main- tenance and CO* concentration in a swarm cluster of honey bees. Apis ~re~~jp~~er~. Camp. ~;oc/~~~~. P~~~~~~)~. 55A, 169-172.

PEARSON 0. P. (1960) The oxygen consumption and bio- energetics of harvest mice. Physiol. Zool. 33, 152-l 60.

PINSHOW B., FEDAK M. A., BALES D. R. & SCHMIDT- NIELSEN K. (1976) Energy expenditure for thermoregula- tion and locomotion in emperor penguins. ARI. J. Ph_v- siol. 231, 903-912.

SCHMIDT-NIELSEN K., DAWSON T. J. & CRAWFORLI E. C. (1966) Temperature regulation in the echidna (Tuchp- glosslls ~C~~~Uf~S). J. crli. camp. Phpsiof. 67, 63-71.

SCHOLANI)ER P. F., HOCK R., WALTERS V. & IRVING L. (1950) Adaptation to cold in arctic and tropical mam- mals and birds in relation to body temperature, insula- tion and basal metabolic rate. Biol. Bull. 99, 259-271.

THIGPEN L. W. (1940) Histology of the skin of a normally hairless rodent. f. tvfutizlmal. 21, 4491156.

WHITE F. N. & LASIEWSKI R. C. (1971) Rattlesnake den- ning: theoretical considerations on winter temperatures. J. theor. Riol. 30, 553-557.

WITHERS P. C. (1977) Measurement of vo>. Vco, and eva- porative water loss with a flow-through mask. J. oppi. Physid. 42, 12&i 23.

WITHERS P. C. (1978) Bioenergetics of a “primitive” mam- mal, the Cape golden mole. Sth A.fr. J. Sci. 74, 347-348.