Huddling by rat pups: Group behavioral mechanisms of temperature regulation and energy conservation

Journal of Comparative and Physiological Psychology1978, Vol. 92, No. 2, 231-246

Huddling by Rat Pups:Group Behavioral Mechanisms of Temperature

Regulation and Energy Conservation

Jeffrey R. AlbertsIndiana University

Body heat loss was attenuated and oxygen consumption was reduced by hud-dling in litters of developing rats. Rat pups derive physiological benefits fromhuddling similar to those enjoyed by adult mammals; these findings contrastwith previous characterizations of the altricial rat as poikilothermic. Huddlinginsulates by lessening the exposed body surface area of the participants, thusretarding heat loss and enhancing the efficiency of thermogenesis. Thesephysical mechanisms of the clump are actively regulated by the pups. A novelquantitative measure of huddle size revealed a form of group regulatory be-havior in rat pups whereby the surface expanded and contracted with in-creases and decreases in ambient temperature. The individual basis of thisgroup regulatory activity was investigated by marking individual pups and ob-serving them in huddles by means of time-lapse videography. It was foundthat individual animals circulate through the huddle, frequently exchanginglocations in the group. By studying the huddle positions of an anesthetizedpup and a marked control sibling, dynamics of the pup flow were clarified. Or-dinarily, the direction of movement was actively downward, into the pile; im-mobile pups "floated" on the surface. When the nest temperature was raisedto thermoneutral, the direction of pup flow reversed and an immobile animalsank to the depths of the huddle. Through individual competitive adjustmentsthe huddle behaves as a self-regulating unit which provides warmth and insu-lation to all its active members.

From birth, the rat is a member of a so- huddling is the behavior that leads to thecial group, viz., the litter and mother. In- formation and maintenance of the litter ag-teraction between rat pups initially consists gregate (Alberts, 1978) and, similarly, pro-of huddling behavior. Broadly defined, duces the social clumps of adult rats ob-

served under group-living conditions (Bar-_ _ _ nett, 1963; Calhoun, 1962; Steineger,

Most of this research was supported by GrantMH28355-01 from the National Institute of Mental Huddling is the major behavioral activityHealth to the author. Experiment 1 and portions of of the infant rat. Pups exhibit vigorous andExperiment 3 were submitted as part of a doctoral dis- persistent huddling behavior with a variety

,Mental Health to Byron A. Campbell and a predoctoral olfactory, photic, and several kinds of tactilefellowship from that agency to the author. The oxygen CUCS all play a demonstrable role in the ex-analyzer and laboratory space for Experiment 2 were pression of huddling by rat pups (Alberts,generously made available by Henry D. Prange and 1973- Cosnier 1965)David Robertshaw, Department of Physiology, Indiana mi' ' /• ' , • i • • uj.University. The careful and creative assistance of Brad There «» many functions clumping mightMay made the metabolic experiments possible. I would serve during early life. In the wild, rat pupsparticularly like to thank B. A. Campbell, J. C. Craig, are born and reared in special burrows thatM. I. Friedman, D. J. Howell, B. G. Galef, C. G. Mueller, are defended by the mother against rats and

d A^-^AS^ thdr 8UPPMt other "- (C-lhoun, 1962). Clumping,Requests for reprints should be sent to Jeffrey R. or any behavior that confines the pups to the

Alberts, Department of Psychology, Indiana University, maternal nest, would therefore aid in pro-Bloomington, Indiana 47401. tection from predation and permit the dam

Copyright 1978 by the American Psychological Association, Inc. All rights of reproduction in any form reserved.

231

232 JEFFREY R. ALBERTS

to leave the nest to forage. By keeping thepups together and easy to localize during themother's periodic visitations, huddling mayalso increase the efficiency of parental in-vestments, such as nursing. In adult animalsat least, huddling is widely considered to bean important behavioral adjustment to coldtemperatures (Whittow, 1971).

Despite the prominence of huddling in thelife history of the rat and its broad phyloge-netic representation (see Whittow, 1971,1973), little is known about the developmentor the significance of the behavior. Cosnier(1965) concluded that gregariousness in in-fant rats is primarily a thermotactile re-sponse. Welker (1959) and Jeddi (1970)reached similar conclusions from their ob-servations of neonatal dogs. Cosnier recog-nized the contribution of huddling to ho-meostatic functions in the fragile neonateand suggested that grouping is an importantcondition for survival and early developmentin the altricial rat.

Two of the experiments reported belowwere designed to examine the effect of hud-dling behavior on two measures of physio-logical function in rat pups: body tempera-ture and oxygen consumption. In addition,three experiments are described whichstudied the behavioral dynamics of the groupitself and of the individuals in the groupduring huddling.

Experiment 1

As noted before, huddling among adultanimals is frequently considered to be anadaptive social defense to cold (Whittow,1971). Many species of animals abandonsolitary habits and live in close associationwith conspecifics during the colder months(Hart, 1971). Huddling attenuates convec-tive heat loss by reducing the exposed sur-face area of an animal. Moreover, in furredor feathered forms, the behavior provideseach participant with a localized area ofthicker insulation. In addition, huddlingwith other warm bodies reduces heat loss byconduction to colder surfaces.

The parameters of body temperatureregulation differ between immature rats andadults. For its size, the infant rat producesless and loses more body heat than does the

adult (Taylor, 1960). Thermogenesis byshivering is absent in the infant, and thelimits of metabolic heat production arebelow that of the adult. Heat loss is rapid inpups because they lack insulative fur andsubcutaneous fat and cannot exert controlover vascular flow (Hull, 1973). Moreover,juveniles have a relatively greater heat-losingsurface area than adults.

Together, these morphological and phys-iological attributes leave the rat pup withthermoregulatory capabilities so narrow thatits body temperature tends to approximatethat of the environment (Hahn & Kold-ovsky, 1966). As a result, many writers havetermed the infant rat "poikilothermic,"thereby likening the process of body tem-perature regulation in this small mammal tothat of a reptile. The question can be raised,then, whether huddling can make a quanti-tatively significant contribution to bodytemperature defense to the rat pups whenthey lack the thermogenic and heat-con-serving capabilities of homeothermic adults.Experiment 1 was performed to evaluate theeffect of huddling on rectal temperature inrat pups of various ages.

Method

Subjects. A total of 96 rat pups were tested. Pupswere tested only once each, at 5,10,15, or 20 days of age.All pups were born in the Princeton University colony,descendants of Sprague-Dawley rats from the breedingpopulation, or were derived from adult rats purchasedfrom Camm Research, Inc. Three days after birth (Day0), litters were uniformly reduced to eight pups each andotherwise left undisturbed with the mother in standardplastic maternity cages, Purina Laboratory Chow andwater were continuously available. Twenty-four pupswere used in each age group.

Procedure. Colonic temperatures were measured tothe nearest .10 °C with a Shulteis mercury thermome-ter. The bulb of the thermometer was sufficiently smallto be accommodated by the youngest pups and was in-serted to a depth of 10 mm. The thermometer's rapidrise time (hence short insertion period) was advanta-geous in minimizing stress-induced hyperthermic re-actions.

Litters were removed from the home cage and initialrectal temperatures were recorded. Pups were markedfor individual identification and then placed in Plexiglascompartments carpeted with home cage shavings andtopped with hardware cloth. Four pups from each litterwere housed singly (isolates) and four pups were kepttogether in a single compartment (huddlers). Temper-ature measurements were made hourly on each pup forthe next 4 hr.

GROUP REGULATORY BEHAVIOR IN DEVELOPING RATS 233

36 - 5-DAYS-OLD

34

~ 32o

tucc

cc1UQ.

UJt-_J

30

28

26

36-

34-

32-

30-

-10-DAYS-OLD

- 15-DAYS-OLD

O—O Single pups

Muddles of 4 pups

2 3 4 0 1 2

HOURS OUT OF NEST

Figure 1. Rectal temperatures of rat pups at 24 °C inhuddles of four siblings (closed circles) or alone (opencircles). (The n = 12 for each condition at each age.)

The ambient temperature ranged from 23 to 24 °C.Relative humidity was not controlled but was monitoredwith a hygrometer and ranged from 40% to 60%.

Results and Discussion

The initial colonic temperatures of thepups did not differ, but after 4 hr out of thenest, the rectal temperatures of the isolated5-, 10-, and 15-day-old pups were signifi-cantly below those of the littermates left inclumps of four (Mann-Whitney U test,two-tailed, ps = .000, .038, .006, respective-ly). The colonic temperatures of 20-day-oldisolates, on the other hand, did not differsignificantly from their huddling siblingsafter 4 hr (p = .253). Figure 1 depicts themedian rectal temperatures of the pups foreach hour of the experiment.

It can be seen in Figure 1 that huddlingattenuated the rate of temperature loss in anage-related manner. Significant (p < .05)differences in rectal temperature betweengrouped and singly housed pups were foundafter the first hour out of the nest in the 5-and 10-day-olds. Fifteen-day-old isolates,however, did not have significantly lower

colonic temperatures until the third hour ofthe experiment. The 20-day-old pups (boththe singles and huddlers) successfully de-fended their initial rectal temperatures forthe entire 4-hr duration of the test period.

The 15- and 20-day-old pups did not showthe same dramatic body temperature de-creases as did the younger subjects. Thesedata are entirely consistent with generaldescriptions of the ontogeny of thermo-regulation in altricial mammals (Barnett &Mount, 1971; Hart, 1971; McCance, 1959)and with Cosnier's (1965) findings on surfacetemperature changes in isolated and groupedpups. During development, heat productionability increases, heat loss decreases withincreasing size, and the insulation value ofthe pelage improves. The temperaturechallenge used here was probably not suffi-ciently severe to see a robust huddling effectin the older pups. In a pilot study (Alberts,1973, unpublished), however, huddling in avery cold (10 °C) environment enabled 15-and 20-day-olds to defend their colonictemperatures more effectively than singletonlittermates.

Huddling appears to provide the rat pupwith an effective behavioral means of re-ducing loss of body heat and thus combatingcold challenge. It is not possible to extendinterpretation of these data to more naturalconditions or even to estimate the extent towhich huddling can augment temperatureregulation in the litter situation. The con-ditions used in this experiment were lessthan optimal for the pups; very small clumpswere used and the insulation of the nest waseliminated. Nevertheless, even the neonatesderived marked thermoregulatory benefitsfrom huddling with small numbers of otherfur less, rapidly cooling siblings.

Experiment 2

Rectal temperature is a useful but limitedmeasure of temperature regulation. Al-though rectal temperature is often inter-preted as an "average" body temperature, itis not representative of the organism as awhole and remains a regional temperaturemeasurement. Moreover, while rectal tem-perature may yield information on the con-sequences of an organism's thermal re-


sponse, it does not reflect the strategy used.Body temperature is the difference betweenthe amounts of heat produced and heat dis-sipated. Two animals may manifest equiv-alent body temperatures, but one may belosing heat more rapidly and therefore exerta greater metabolic effort to maintain thesame colonic temperature.

A more direct measure of thermogeniceffort is found in metabolic rate. Oxygenconsumption (VC^: volume of oxygen con-sumed/body weight/time) is a standardtechnique for determination of metabolicrate and can be used to quantify thermoge-nesis. Homeotherms increase their metabolicrate in cold temperatures; thus, their rate ofVC>2 is lower in warm temperatures, reachingits minimum in the range defined as the zoneof thermoneutrality. In true poikilotherms,however, metabolism varies directly withambient temperature and there is no zone ofthermoneutrality.

Because the rectal temperature of a juve-nile rodent is likely to decrease precipitouslyin a cold environment, the pup is oftentermed "poikilothermic" (Fairfield, 1948;Fowler & Kellogg, 1975). Taylor (1960)showed, however, that neonatal rats increasetheir metabolic heat production by 100% ifthe temperature challenge is not too severeto combat. Classifying the neonatal rat as"poikilothermic" is not accurate and can beseriously misleading (Hull, 1973).

Huddling behavior has been implicated asa behavioral means of reducing metabolicexpenditure in adult homeotherms. Adultmice permitted to huddle in cold environ-ments survive longer than singly housedmice (King & Connon, 1955; Sealander,1952), presumably because huddling reducesmetabolic expenditures (Howard, 1951;Pearson, 1960). Prychodko (1958) calculatedthat at —3 °C, the nutritive energy require-ment of a mouse in a huddle of five is re-duced by almost 30%, a metabolic responseequivalent to that of raising the ambienttemperature 11 °C.

The present experiment investigatedwhether huddling could conserve metabolicenergy in rat pups. It was possible that theresults of the previous rectal temperatureexperiment told only part of the story; eventhe "poikilothermic" neonates, whose body

temperatures dropped dramatically, mayhave been making metabolic responses notapparent from their colonic temperatures.Past research, using various measures ofmetabolic rate, has been inconclusive. Taylor(1960) found only meager differences inoxygen consumption by grouped and iso-lated pups, but his animals were tested atthermoneutral or very cold (10 °C) temper-atures. Cosnier (1965) reported that hud-dling reduced oxygen consumption by ratpups aged 1-20 days, at either 21 or 32 °C.More recently, Bryant and Hails (1975)suggested that "before homeothermy is fullydeveloped" in mice (Mus musculus), hud-dling may actually increase oxygen con-sumption. Their measures, however, ap-peared to be confounded by movement ar-tifacts in the respirometer. The question ofthe metabolic role of huddling in juvenilestherefore remains open. The study reportedbelow utilized a procedure and an apparatuswell suited to accurate measures of metabolicresponses in neonatal rats.

Method

Subjects. A total of 96 Sprague-Dawley pups (12litters) were used in this experiment. The litters wereborn in the Indiana University colony and were derivedfrom a line of rats initially purchased from LaboratorySupply Inc., Indianapolis. The breeding and rearingprocedures were as described in the previous experi-ment.

Apparatus. The respiratory chambers were con-structed from clean commercial paint cans. These in-expensive containers are sturdy but maleable, are watertight, and with the use of stopcock grease, can be re-peatedly opened and resealed. Hardware cloth baskets(1 cm mesh) approximately 2 cm smaller than the in-ternal dimensions of the cans were constructed to pre-vent the animals from directly contacting the walls ofthe chamber. Legs, .5 cm in length, prevented contactwith the floor. There were four such chambers madefrom 1-qt. (.95-1) cans and one chamber made from a1-gal. (.004-m3) can. Each lid was fit with two pipes (3.5mm ID) which served for air inlet and exhaust. Air en-tered near the base of the chamber through the longerpipe and was exhausted from the shorter effluent nearthe ceiling. A mercury thermometer was also sealed intoeach chamber lid.

The small, 1-qt. chambers were used to test singlepups and huddles of two. Huddles of four and eight wererun in the larger, 1-gal. can. Another respiratorychamber, adapted from a 5-gal. (.02-m3) commercialaquarium was fit with a Plexiglas lid and arranged in-ternally in a manner similar to the cans described above.This chamber permitted subjects to be directly ob-served. The results of this experiment using the two

GROUP REGULATORY BEHAVIOR IN DEVELOPING RATS

Air f low rate measuring system

235

Pump

Overflow valve

Regulator

OxygenAnalyzer Chart Recorder

~A IV — ~-> :

Water bathcontaining

respiratory chamber(s)

Telethermometer

Figure 2. System for analysis of oxygen consumption.

types of chambers were the same, so no distinction ismade between them in the discussion below. The testingchambers were submerged to within 5 cm of their topin a temperature-controlled water bath. Air tempera-ture in the chambers was either 28 or 30 °C, as describedbelow.

The system used to measure oxygen consumption isdepicted in Figure 2. A Neptune Dyna-pump drew at-mospheric air through the respiratory chambers andinto a damping chamber which also served as a manifoldand water trap. Total flow rate, divided among the fiverespiratory chambers or through the larger glasschamber, was maintained at 1,200 ml/min. Still undernegative pressure, the air passed through a column ofdessicant (Drierite) and C02 absorbant (soda lime).

The air flow rate-measuring system consisted of a250-ml displacement flowmeter (Vol-U-Meter, BrooksInstrument Div., Emerson Electronics), a water ma-nometer, needle valves, and a source of compressed air.The calibration apparatus was connected to the gas-analyzing system by a T valve, as shown in Figure 2.

Air left the pump under positive pressure. An ad-justable overflow valve was open on-line so that a con-stant flow of air, regulated at 5 in. (12.7 cm) of water(Matheson regulator) entered an Applied Electronicsoxygen analyzer (Model S-3A, Sunnyvale, California),accurate to .001%. In addition to a digital display ofpercentage of Og entering the analyzer, the output wastranscribed continuously onto a Linear Instruments(Irvine, California) stereo chart recorder. The full penexcursion of 26 cm, corresponding to 1.0% oxygen,yielded a high degree of sensitivity. A second channelon the chart recorder monitored the bath temperatureby means of a Yellow Springs telethermometer.

Procedure. The entire system was calibrated at the

beginning of each experimental session. Air pressureand flow rate were adjusted with the displacementflowmeter and water manometer. After flushing theentire system with atmospheric air for 15 min, theoxygen analyzer and chart recorder were calibrated.

Rat pups were weighed individually and then placedin a single, large chamber as an intact litter (n = 8).When 5-day-old huddles were tested, the pups wereplaced in a glass finger bowl (11X5 cm). Pilot data anddirect observation indicated that these neonates haddifficulty maintaining a unified clump in the apparatuswithout a concave substrate. After a 15-min habitationperiod, V02 was recorded continuously for 75 min. Thelitter was then subdivided into two huddles of four pupseach and tested similarly. Half of the litter remainedwith the dam while the other four pups were beingtested. By subdividing again, clumps of two pups eachand singletons were tested. Removable perforatedpartitions in the 5-gal. chamber allowed simultaneoustesting of the entire litter, with the pups separated orin clumps of two, four, and eight. Litters were testedboth in descending and ascending order of group size.

The 10-, 15-, and 20-day-olds were tested at 28 °C, atemperature below thermoneutral (Taylor, 1960) butnot stressful to the pups. Five-day-olds performed er-ratically at this temperature in pilot studies, probablybecause the neonate's thermoneutral zone is somewhathigher (Taylor, 1960). Therefore, the ambient temper-ature of the test chamber was increased to 30 for the5-day-olds.

Upon completion of the last test, the empty chamberswere resealed, and for 15 min the system was againflushed with atmospheric air and the calibrations werechecked. As is typical in such long-term metabolicmeasures, rate of oxygen consumption was calculated


Table 1Mean Rate of Oxygen Consumption (in ml/glhr) of Developing Rats Tested Alone and inHuddles of Littermates of Various Numbers

Age(in days)

555599

101115151516

No.Single

3.382.663.053.503.463.703.462.712.512.152.372.23

of pups2

2.762.562.863.41——

2.59———

1.74—

in group4

2.312.352.102.772.403.102.451.901.841.901.671.69

8

2.242.251.712.212.102.492.101.741.701.621.481.47

Note. Five-day-olds were tested at an ambient temperature of30 °C, and the 9- to 15-day-olds were tested in a chamber at 28°C. Not all litters were tested with only two pups/huddle, butin each of the three age groups shown, some litters were testedin ascending and descending order of huddle size. Pups in eachlitter were tested alone and in groups of various sizes. Thus, themean given is for eight pups in each case.

from an extended period of stability. A 20-min plateauperiod from the last half of each session was used, and10 VO2 readings, 2 min apart, were taken from the chartrecords. Values were corrected to standard temperatureand pressure and are expressed below as milliliters ofoxygen consumed/gram/hour.

Results

In this experiment, huddling significantlyreduced oxygen consumption by rat pups.The magnitude of the metabolic savings wasrelated to the size of the huddle. The meanvolume of oxygen consumed (V02) by indi-vidual pups was 2.93 ml/g/hr across all agestested. In huddles of eight siblings, VC>2 bythe same pups averaged 1.93 ml/g/hr,yielding an overall mean reduction of 33.8%.The largest reduction in V02 observed,43.9%, was in a litter of 5-day-olds. Table 1presents a summary of the results obtainedin this experiment. It can be seen in Table 1that the 12 litters studied in the present ex-periment displayed a consistent pattern ofresults. Maximum rates of oxygen con-sumption were obtained from singletons, andlower VC>2 was associated with huddling, asa function of the number of pups per clump.Although the absolute rates of oxygen con-sumption differed considerably among some

of the litters tested, the within-litters vari-ability in VC>2 was negligible, ranging from.01% to .06% O2.

Percentage of reduction in oxygen con-sumption, by comparison of V02 by single-tons with that of the same pups clumpedtogether in groups of two, four, or eight, en-ables one to evaluate the proportional re-duction in metabolic rate produced by hud-dling. Again, a highly consistent pattern ofresults was obtained. For the three agegroups tested: 4-5-, 9-11-, and 15-16-day-olds, rate of oxygen consumption averaged32.5%, 36.8%, and 32.2% less, respectively, inpups huddled in groups of eight than whenthey were separately measured. Huddled ingroups of four, pups in these age groupsconsumed 23.9%, 26.5%, and 23.0% lessoxygen, respectively, than their rates takensingly. Relatively few huddles of two pupswere tested, as can be seen in Table 1, but itwas apparent that even two pups, thesmallest huddle possible, decreased VC>2.

It should be recalled that the ambienttemperature of the testing environment wasvaried for the different age groups (seeMethod) in order to conform to the shift inthe level of the juveniles' thermoneutral zoneduring early development. Thus, additionalage-wise comparisons of metabolic savingswould not be profitable here.

Discussion

The results of the present experimentshow that huddling provides the rat pupwith an efficient means of reducing the ex-penditure of metabolic energy (oxygen con-sumption). This phenomenon was found inrat pups ranging from 5 to 16 days of age.The magnitude of the metabolic savingsmeasured in the present experiment wasconsiderable, averaging about 34% for a litterof eight pups. Figure 3 shows the pups' oxy-gen consumption when alone and in litter-mate groups of different numbers.

The effect of huddling on metabolic ratefound here is the opposite of what would bepredicted if the rat pup was truly poikilo-thermic. Metabolic rate in a poikilothermicanimal is a direct function of temperature;VC>2 increases with increasing body tem-perature. It was shown in Experiment 1 that


zgi-Q.5

Oo

Zu

O—O 5-doys-oldi-days-old

2 4

NUMBER OF PUPS IN CHAMBER

Figure 3. Rate of oxygen consumption by rat pupa ofvarious ages tested alone and in huddles of different

huddling reduces heat loss and thus keepspups warmer. In the present experiment(Figure 3) pups had lower metabolic rateswhen they were warmer in huddles thanwhen heat loss was increased by placing theanimals in smaller groups or in isolation.

Overall, the findings of this study are inbasic agreement with the existing data on theeffects of grouping on oxygen consumptionin rats (Cosnier, 1965; Taylor, 1960). How-ever, unlike earlier research, the presentexperiment tested huddles of various sizesand, in doing so, strengthened the assertionthat the resultant metabolic effects can beattributed to huddling per se. Studies thatsimply compare grouped and single animalsmay obtain results reflecting relativelynonspecific differences, such as those causedby isolation distress (see Randall & Camp-bell, 1976), rather than processes charac-teristic of grouping. The group-size-relatedreduction in oxygen consumption found herestrongly suggests that the phenomenon isrelated to some specific aspect of hud-dling.

The present findings show more consis-tent metabolic reductions across ages thandid the previous studies. In contrast to theearlier studies, the ambient temperaturesused here were only moderately challengingto the pups, i.e., 2-4 °C below thermoneutral.In many past experiments, rats have beentested at severely cold temperatures, whichmay rapidly induce torpor, or in thermo-

neutral environments, which, by definition,minimize metabolic responses.

The findings of Experiments 1 and 2 canbe explained by the consequences ofclumping on surface area. Rat pups, likeother physical entities, lose heat at a rateproportional to their surface/volume ratio.Huddling reduces the exposed body surfacearea of each pup in the group, thereby de-creasing the surface/volume ratio and heatloss. The huddle, in effect, increases the bodysize of the animals in it. With reduced heatloss accompanying increased size, metabolicrate can decrease, as it does for homeothermsin general (Schmidt-Nielson, 1975).

Experiment 3

Experiments 1 and 2 were concerned withthe physiological consequences of huddlingbut did not address the behavioral mecha-nisms by which these results are obtained.An important question therefore arises: Arethe thermal and metabolic consequences ofhuddling merely the passive effects of astereotyped tendency to aggregate, or canhuddling be viewed as an active regulatorybehavior on the part of the group? The sig-nificance of this question involves a funda-mental bias in our view of the altricial infantand the nature of postnatal development.

A commonly held view of the infant as ahelpless, reliant, and passive organism em-phasizes the asymmetry of the parent-off-spring relation and focuses on the juveniles'needs and incapabilities. This conceptualframework has spawned a body of data that,not surprisingly, demonstrates the relativeinability of juveniles to solve problemscommon to adults. Much of the early workon the ontogeny of temperature regulation,for instance, consists of demonstrations thatinfant rodents, tested in the same manner astheir adult counterparts, suffer relativelylarge decreases in body temperature andmanifest reduced oxygen consumption(Conklin & Heggeness, 1971). Consequently,we have a persisting tradition of inaccuratereference to the neonate as "poikilother-mic."

Experiment 3 sought to examine the ex-tent to which huddling might be an activeregulatory behavior rather than a fixed re-


sponse resulting in a heap of otherwise pas-sive pups. Such an empirical demonstrationwould warrant a shift in emphasis andawareness, to focus more on the regulatoryabilities of the infant and the functional or-ganization of neonatal behavior. It was rea-soned that if huddling is modulated by en-vironmental temperature, then one wouldexpect a clump to form as a body of minimalsurface area at low ambient temperaturesand, with increasing heat, the surface area ofthe clump would increase. Previous observ-ers have noted that immature rats and micegroup and disband as a function of temper-ature (Cairns, 1972; Cosnier, 1965). Thepresent experiment describes a method usedto investigate group regulatory dynamics ofhuddling on the form of the clump.

Method

Subjects. A total of 48 pups were tested. Threegroups of four littermates each were tested at 5,10,15,and 20 days of age. Animals were bred and maintainedas described in the experiments above.

Procedure. To examine the surface area of theclump as a function of ambient temperature requireda method of measurement of clump size (surface area)and a means of changing the animals' environmentaltemperature.

Surface area measurements of clumps of pups wereobtained as follows: A tripod-mounted 35-mm camerawas positioned directly above the animals' chamber(described below). Photographic transparencies weretaken of subjects when they were clumped and whenthey were not in contact. Tracings of the pups were latermade from projections of the slides from a fixed dis-tance. The circumference of the clumps was measuredwith a standard commercial hodometer (map measurer).For purposes of correction for age-related differencesin pup size, distance of camera to cage, and enlargementduring projection, 6-10 measures of individual pupswere made from each roll of film (all parameters wereconstant within each roll). An average circumference ofan individual pup was derived from the tracings of 6-10individuals. This average pup circumference was arbi-trarily assigned the value 1.0. The total circumferenceof the four individual pups in an observation cage whichwere not in contact with one another was therefore4.0.

For measurement of the exposed surface area ofhuddles, tracings were made of clumps which eliminatedall areas of overlap between animals. An example of thisis shown in Figure 4. The left portion of the figure is atracing of a clump of 10-day-old pups, drawn to repre-sent the actual appearance of the pups in contact anddepicting their postures in relation to one another. Theright portion of the figure illustrates the exposed surfacearea of this clump. Note that in the latter tracing, por-tions of a pup's body that are in contact with another

Figure 4. Conversion of a tracing of a huddle into anoutline reflecting the exposed huddle "surface." (Por-tions of pups contacting or overlapping with other pupswere eliminated in the conversion, leaving an outline ofexposed pup surface.)

pup have been eliminated so that the drawing repre-sents the exposed surface area of the group viewed fromabove. The linear circumference in this tracing, cor-rected for average pup size, represents an estimate oftotal exposed surface area of the clump. With thismeasure, clump surface could theoretically range invalue from a minimum of 1.0 (if pups were stackedperfectly on top of each other when photographed fromabove) to a maximum of 4.0 (if a clump of four pups wascompletely dispersed with no pup touching another).

Control of ambient temperature in the test cage wasaccomplished by changing the temperature of a waterbath in which the observation chamber was partlysubmerged. Changes in chamber temperature effectedin this manner are gradual and uniform.

Molded opaque plastic observation cages wereweighted and partially immersed in a commercial 20-gal. aquarium. The water level reached to within ap-proximately 2 mm of the top of the cage. The cage wascovered with a perforated sheet of clear Plexiglas. Thecage floor was insulated from the water below it by alayer of corrugated paper covered by woodshavingstaken from the subject's home cage.

Cold ambient temperatures were produced by addingice to the aquarium water; warm chamber temperatureswere achieved by removing ice, siphoning water, andreplacing it with sufficient quantities of warm or hotwater to reach desired temperatures in the observationcage. Cage temperature was monitored with a YellowSprings telethermometer (Model 73TD) and an airtemperature probe.

The experiment involved placing groups of four lit-termate's into the partially submerged observation cage.After a 15-min habituation period the ambient tem-perature of the cage was recorded, and a photograph wastaken from above the cage. Thereafter, every 15 min andfor the duration of the experiment, a temperaturereading and a photograph were taken. Following each10-min period, the water temperature was altered, asdescribed above, so that the chamber temperature in-


3.0

15-OAVS-OUO _ 20-DAYS-OLD

20-22 26-28 32-34 3840 34-32 23-26 22-20 20-22 26-26 32-34 36-40 34-32 26-26 22-20

AMBIENT TEMPERATURE PC)

Figure 5. Regulation of the exposed huddle "surface" area by rat pups of various ages. (Each continuous line depictsthe average exposed surface area of a clump of four siblings at various points during a cycle of temperature change,shown along the abscissa. The exposed surface of a clump ranged from 1.0 to 4.0, by the method of conversiondescribed in the text.)

creased until it reached 40 °C, at which time the watertemperature was decreased gradually until the ambienttemperature in the cage was 20 °C. The entire 20-40-20°C cycle lasted 1.5-2 hr. Some of the youngest subjectswere tested in a somewhat narrower, and potentially lessdebilitating, temperature cycle, beginning around 26 °Cand reaching only 37 °C before descending.


The most conspicuous and significantfinding in this experiment was that groupsof rat pups interacted so as to adjust the totalexposed surface area of their clumps in cor-respondence to the ambient temperature. Asthe temperature of the animal chamber roseand fell, the total exposed surface area of thehuddles increased and decreased. The vol-ume or mass of each huddle remained con-stant, of course, so that the surface/volumeratio of the clumps was changed in accor-dance with the ambient temperature.

Figure 5 illustrates the results, showingthe alterations in "clump surface area" as afunction of ambient temperature in huddlesof 5-, 10-, 15-, and 20-day-old pups. Eachsingle continuous line represents the be-havior of a single clump during the temper-ature cycle. The graphs depict the clump

sizes at 3.0 °C intervals. Therefore, most ofthe points on the graphs actually representthe average of several photographs takenwhen the chamber temperature was withinthe range specified on the abscissa.

The method of measurement utilized inthe present study appears to be a usefulmeans of studying a novel dimension of thebehavior of groups. The technique revealsthat young rats interact so as to regulate thesize of the clump. The rate of temperaturechange varied considerably in this experi-ment due to the relatively crude method ofheating and cooling the water bath; some ofthe differences between groups were un-doubtedly produced by inconsistencies in thetemperature cycle.

The positive correlation of huddle surfaceto ambient temperature can be seen at eachage. It should be noted that all of the clumpsstudied were observed to disband completelyat the peak temperatures and eventuallyreform as the temperature of the chamberdeclined. Despite some of the variations seenin Figure 5, it is clear that rat pups 5 days ofage and older are capable of adjustingthemselves and the clump to conform tochanges in environmental temperature.


It is likely that this form of "group regu-latory behavior" can regulate heat loss and,correspondingly, modulate the pups' meta-bolic rate. This would suggest that thephysiological consequences of huddling arenot passively derived from the clump. In-stead, it suggests that the behavior is awell-tuned, integrated activity which regu-lates the group as a whole, resulting in ther-moregulatory and metabolic benefits sharedby the group members.

Experiment 4

A huddle of infant rats appears to alter itsform and structure in an adaptive fashion.By means of behavioral interaction, rat pupsregulate the exposed surface area of thehuddle in response to the ambient temper-ature. Although the litter aggregate can beprofitably viewed as a regulative unit, it isnecessary to examine the behavior of theindividual pups that constitute the huddleto understand how group regulation mightbe accomplished.

Not all positions in the huddle provideequivalent insulation or allow for rapid heatdissipation. At cool ambient temperatures,a place in the interior of the clump, mini-mizing body exposure, for instance, would beadvantageous for the individual; a peripheralposition on the huddle surface would be theleast effective site from which to derive theinsulative benefits of clumping. Conversely,dissipation of body heat by convection orevaporation would be more efficient on thesurface of the huddle than in the core. If thepups maintained constant positions in thelitter, then the observed temperature-de-pendent changes in huddle size would havelittle or no effect on these pups located onthe periphery of the group. Informal obser-vation, however, indicated that pups dochange their position in the huddle. Thestudy described below was designed to ex-amine some of the individual behavioraldynamics within the huddle that couldcontribute to both group and individualregulation.

Method

Subjects. A total of 36 pups were used. The rats werereared as described above and were tested in six groups

of six littermates each when they were 10-12 days ofage.

Apparatus. Huddles of rat pups were studied in abowl-shaped nest. The nest was made from a polypro-pylene funnel, cut horizontally to form a truncated cone,11.5 cm high (top diameter, 15.7 cm; bottom diameter,6.5 cm) with a 60° sloping wall. The modified funnel fitsnuggly into a 24-cm-high Plexiglas cylinder. The bot-tom of the nest was covered with woodshavings from thesubject's home cage.

A closed-circuit television camera was positionedabove the nest and a close-up lens provided a clear anddetailed picture of the pups in the nest; normally twoor three pups were fully visible from above. Test sessionswere recorded on a time-lapse videotape recorder (Hi-tachi-Shibaden Model 512-U) and scored during rapidplayback. The time-lapse video system and scoringapparatus are described in detail elsewhere (Alberts,1978).

Procedure. Six pups were removed from their homecage immediately before testing. Two of the pups, se-lected randomly, were marked for individual identifi-cation. One pup was marked with a solid stripe (about4 mm wide), extending along the dorsal midline fromthe level of the shoulders to the hips. The second pupwas marked similarly, with a dashed line covering thesame body region. The entire group of pups was placedinto the nest of the testing apparatus, and after a 20-minhabituation period, the activity of the clump was re-corded by time-lapse videotape for 2 hr. Recordingswere made at 5 fields/sec and were viewed for scoringat 60 fields/sec (12:1 record/playback ratio).

The videotapes were scored by viewing the experi-ment during high-speed playback and measuring thetime-spent-exposed on the clump by each of the markedpups. The identification marks were used to define thepups' exposure; when the entire dorsal marking wasvisible, the experimenter activated an electromechanicaltimer and counter. Thus, time-spent-exposed refers toonly the dorsal mark on the pup. Body regions could beexposed or buried in the clump without being scored assuch. The electromechanical scoring apparatus wasprogrammed to sum the total duration of exposure andthe number of appearances on the huddle surface forsuccessive 7.5-min intervals. Some of the videotapeswere scored by two independent observers to assess thereliability of the measures. Values obtained in thesetests differed less than 5% between observers. In addi-tion, all the videotapes were viewed again and scoredwith an Esterline-Angus event recorder which provideda record of the number and the duration of each boutof exposure.


The time-lapse record revealed that thehuddle is a restless, almost continuouslyactive mass of bodies. Pups appeared anddisappeared from view on the surface of thehuddle. Figure 6 is a composite of drawingsmade of a clump of six 10-day-olds in a glassfunnel, depicting the movements studied in


100

80

Figure 6. Four views of the flow of pups through thehuddle. (The drawings were taken from photographictransparencies of a clump of six 10-day-old siblings ina glass funnel. The arrows depict the movements thatwere analyzed in Experiments 4 and 5.)

this experiment. The drawings are fromphotographs taken at various angles, modi-fied with arrows to emphasize the dynamicnature of the pile. Experimental data werecollected from a video camera located di-rectly above the group. From the individu-ally marked pups in the pile, quantitativedata were derived to describe these move-ments.

Figure 7 shows four representative indi-vidual pups studied in this experiment. Eachgraph in Figure 7 depicts the record of asingle, individually marked animal. Eachpoint shows the percentages of time duringconsecutive portions of the test that the dyedregion on the subject's back was visible. Itcan be seen in the figure that the pups peri-odically emerged from and disappeared intothe huddle. The flow of pups in and out ofthe depths of the clump appeared as con-vection currents of bodies circulatingthrough the group.

Note that the individual pups describedin Figure 7 could remain visible or obscuredwhile other pups in the same clump changeposition. The individual pups labeled a andc in Figure 7, in fact, were littermates in thesame clump. By superimposing panels 7aand 7c one can visualize part of the com-plexity and rapidity of behavioral dynamicswithin the huddle. Positions in the clump

CLXUJ

LU

p

O

100

HOURS

Figure 7. Time spent exposed by individual pups on thesurface of the huddle during successive 7.5-min intervalsof 2-hr observations. (The graphs depict the cyclicpattern of appearance and disappearance of pups on thehuddle surface. Pups in panels a and c were littermates;their patterns of movement can be compared by su-perimposing those panels in the figure.)

changed even more frequently than is ap-parent in Figure 7 because those graphs showcumulative time exposed per 7.5-min block.The average number of bouts of exposure onthe huddle surface was 25/2-hr session(range = 18-47).

The rapid cycles of relocation in the hud-dle suggest that the pups' behavior, in ad-dition to regulating clump size (exposedsurface area), enables them to share both thecosts of providing insulation and the benefitsof the thermal protection.

Experiment 5

Individual rat pups are observed to appearand disappear periodically among theirsiblings in the huddle. The movements of thelitter establish "convection currents" ofbodies in which individual pups flow throughthe clump. The purpose of Experiment 5 wasto examine more precisely the organizationof the behavioral interchange in the hud-dle.

There are several kinds of responses thatcould mediate the exchange of position in thehuddle. Movement of pups could be pre-dominantly downward, into the clump. Theapparent flow of bodies would be actuated


by peripherally located animals diving andburrowing under the other pups in the group.Alternatively, the phenomenon of periodicappearance of pups could be created by an-imals emerging upward from the group andclimbing upon siblings, thereby providinginsulation until they were covered by a sim-ilar response made by a pup from below.Finally, a third possibility is that there is acomplex form of cooperation among thegroup members of the clump, based on asharing of the benefits and demands ofhuddling. Althqugh it is difficult to specifythe behavioral mechanisms that could or-chestrate a cooperative group process, thishypothesis should not be immediately re-jected.

Experiment 5 consisted of two manipu-lative studies using the basic observationaltechnique described in the previous experi-ment. The fate of an anesthetized rat pup ina clump of active siblings was studied underdifferent temperature conditions. The gen-eral hypothesis for this experiment was thatthe flow of pups in the huddle is based pri-marily on regulatory adjustments made byindividual pups for themselves. If true, thenat cool temperatures individual pups shouldmove more deeply into the group. An im-mobilized pup would therefore remain on thesurface of the clump. Conversely, this hy-pothesis would predict that under warmambient conditions, pups would tend tomaximize their own exposed surface area,creating an upward movement. Conse-quently, an immobilized pup would remainbeneath the active group members.

Method

Subjects. A total of 12 litters were used. Rats were10-12 days of age when tested. Litters were reared in theIndiana University colony as described above, but onlysix pups from each litter were tested.

Procedure. Six robust littermates were removedsimultaneously from their home cage. Two pups, se-lected randomly, were marked for individual identifi-cation. The markings were similar to those used in Ex-periment 4 except that additional lines were placedlaterally, equal in length and parallel to the dorsal mark.The lateral marks enabled the experimenter to use thesame criteria as in the previous experiment for definingexposure on the clump, even when a pup rolled onto itsside and obscured the dorsal mark.

All six pups, including the marked subjects, were

placed in the bowl-shaped nest described in Experiment4. The nest was contained in a tall, waterproof, Plexiglascylinder which was submerged in a water bath to a level10 cm above the nest. The temperature of the waterprovided an ambient temperature of either 24 ± .5 °C(cool) or 36 ± 2 °C (warm) in the nest. Nest temperaturewas monitored by a Yellow Springs telethermometer(Model 41 TD) and an air probe.

After a 15-min habituation period the two markedpups were removed from the clump for injections. Onepup was anesthetized with Equi-Thesin (.023 ml/kg, ip)and the other pup was given a control injection of iso-tonic saline, equal in volume to the anesthetic injection.After the anesthetic had taken effect, the two markedpups were placed together on top of the clump of lit-termates.

Time-lapse videotape recordings were made contin-uously for the next 2 hr. The record/playback ratio usedwas 12:1 as described in the previous experiment. Vi-deotapes were viewed later, during rapid playback, andthe duration of exposure of each of the marked animalswas measured for eight successive intervals (7.5 mineach) of the entire 2-hr test.


In every case, the immobilized rat pupreturned to a "cool," 24 °C nest did not showthe typical pattern of periodic disappearanceand reappearance in the huddle. Instead, theanesthetized pup "floated" on the surface ofthe clump, remaining visible for most of thethe 2-hr test. In contrast, the saline-injectedlittermate control resumed circulatingthrough the group. The left panel of Figure8 presents these results, showing that theimmobilized pups were visible for a mean of86% of the 2-hr test whereas the controlpups, tested simultaneously, were visible foronly 32% of the test period.

Immobile pups returned to their litter ina warm (36 °C) surround rapidly sank to thebottom of the pile where they remained, outof view, for most of the test session. Again,the littermate control pups intermittentlydisappeared and reappeared. The rightpanel of Figure 8 shows the clear differencebetween the amount of time spent on thehuddle surface by active and immobile pups.In a warm nest, anesthetized pups sank intothe pile and were seen for a mean of only 28%of the test. Active control pups in the samelitter were exposed an average of 62% of thesession. The differences between the anes-thetized and control pups were statisticallysignificant in both studies (ps < .05, signtest, two-tailed).


uiQQ

QUI«O0-X

UJOoruiQ.

100

20

Figure 8. Median percentage of time spent exposed ina huddle by an anesthetized pup and a littermate con-trol in nests maintained at cool and warm temperatures(n - 6 for each bar).

Figure 9 is a more detailed view of theoutcome of this experiment, showing thepatterns of movement observed in repre-sentative litters. The dashed lines in eachpanel of the figure depict the behavior of theactive control pups in the two conditions. Asin the previous experiment (see Figure 7)these pups periodically moved in and out ofview in the huddle. Their immobilized lit-termates, however, were considerably dif-

ferent. In a cool nest (left panel) the immo-bile subject, like the sibling control, beganthe test session on the surface of the litter. Ascan be seen in Figure 9, the anesthetized pupremained on the surface of the huddle, pro-viding insulation to the circulating litter butreceiving virtually none. The immobile pupdepicted in the left panel of Figure 9, likemost of the anesthetized animals tested,eventually rolled onto its side and drifted tothe wall of the nest. This pup disappearedfrom view briefly, for a total of about 4.2 minof the 2-hr test, before it was again lifted tothe surface by the movements of its litter-mates.

The fate of an immobilized pup in a hud-dle at a cool ambient temperature suggeststhat the direction of the flow of pups is ac-tively downward and that the appearance ofa pup on the surface of the group is the resultof displacement from the core. The anes-thetized "floater" is maintained on the pe-riphery of the clump by the lifting andpushing of the litter.

In contrast, the immobile pup in a warmlitter environment rapidly sank to the bot-tom of the pile. The right panel of Figure 9compares the performance of the experi-mental and control pups after their return tothe warm nest. The experimental pup sankinto the clump and remained obscured fromview by the cycling litter above. The inactivepup represented by the solid line in the rightpanel of Figure 9 briefly became visible when

COO0-XUJ

UIOcruiQ.

10090

so70

60

50

40

30

20

I 0

0

24° C 36° C

0.5 1.0 IJ5 2.0

HOURS

0.5 1.0 1.5 8.0

Figure 9. Examples of the movements in the litter of individual immobile and control pups at cool and warm nesttemperatures. (Each point shows the percentage of time a pup's identification mark was visible during successive7.5-min portions of the 2-hr observation.)


the siblings had almost all oriented along theside of the nest and effectively left theanesthetized pup exposed. Thus, in a warmnest the overall direction of active flow ofbodies is the reverse of that seen in the coolnest. Pup flow is upward; the rats emergeand pile onto the littermates. An inactivemember of the huddle thus becomes a"sinker" and disappears from view.

It is interesting to note that even withinthe confines of the funnel-type nest usedhere, pups increased the total exposed sur-face area of the group. This increase is ap-parent if the time-spent-exposed by rats ina funnel nest at 24 (Figures 6 and 9, leftpanel) is compared with that by rats in the36 °C nest (Figure 9, right panel).

Welker (1959) and Cosnier (1965) bothobserved altered behavioral responsivenessin neonates as a function of temperature.Likewise, the overall pattern of results foundhere, namely, temperature-dependentchanges in the organization and direction ofthe flow of bodies through a huddle, appearsto be based on rather simple regulatory re-sponses made by individual pups in the lit-ter. Cooler pups burrow downward, into theinsulation of the group. Their movementsdisplace other pups to the periphery. Incontrast, warm pups maximize their ownexposed body surface area and thus activelyascend from beneath other pups to the sur-face, changing the overall direction ofmovement.

General Discussion

Huddling by rat pups functions to reduceheat loss and to conserve the expenditure ofmetabolic energy. In these respects, littersof infant rats resemble many species inwhich adults aggregate seasonally as a socialdefense against cold. Despite their small size,poor insulation, and meager thermogenicability, rat pups nevertheless derive quan-titatively significant thermoregulatorybenefits from huddling.

The metabolic savings engendered byhuddling is undoubtedly of great significancein the energy budget of development. Byhuddling, pups utilize far less metabolicenergy for thermogenesis and can thereforechannel this energy into processes of growth

and development. The sole source of thelitter's nutritive energy is, of course, themother's milk. Energy conservation fromhuddling allows the dam's finite supply ofnutritive energy to contribute more effi-ciently to the weaning of her young, yieldinga reproductive advantage for this behavior.This general interpretation is supported bythe finding that low ambient temperaturesdepress growth rate in young rodents (Bar-nett & Mount, 1971).

The important physiological consequencesof huddling are not simply the result of pilingbodies upon one another. The pup's mode ofactivity in the litter leads to an elegant formof "group regulatory behavior" whereby thehuddle acts as an adjustable unitary body,responding adaptively to changes in ambienttemperature. Arranged loosely in warmtemperatures and tightly cohesive in thecold, the huddle expands and contracts,maximizing and minimizing the heat-dissi-pating surface area of the clump (Experi-ment 3). By doing so, heat loss is significantlyreduced (Experiment 1), which enables thepups to save considerable amounts of met-abolic energy in the processes of thermoge-nesis and body temperature regulation(Experiment 2). Moreover, the pups ex-change positions in the huddle and, in effect,share the costs and benefits of the groupactivity (Experiment 4).

Thus, the huddle behaves as an active,dynamic, regulatory unit. In particular, theresults of Experiments 3-5 showed theadaptive modifiability of the huddle, pro-duced by individual adjustments of thegroup members.

Together, these considerations support aperspective that contrasts sharply with thecharacterization of the altricial infant as ahelpless or entirely dependent creature. Ingroups, infant rats were found to demon-strate exquisite behavioral regulatory abili-ties and to manifest a metabolic strategycharacteristic of homeotherms rather thanthe so-called "primitive" poikilothermicresponse pattern. The findings reported hereunderscore the critical importance of in-cluding the typical ontogenetic milieu in ourstudies of development. For the developingrat the milieu is the huddle; the results canreveal both quantitative and qualitative


capacities that are not apparent in the indi-vidual pup.

References

Alberts, J. R. Huddling by rat pups: Multisensorycontrol of contact behavior. Journal of Comparativeand Physiological Psychology, 1978, 92, 220-230.

Barnett, S. A. A study in behaviour. London: Methuen,1963.

Barnett, S. A., & Mount, L. W. Resistance to cold inmammals. In A. H. Rose (Ed.), Thermobiology. NewYork: Academic Press, 1971.

Bryant, D. M., & Hails, C. J. Mechanisms of heat con-servation in the litters of mice (Mus musculus L.).Comparative Biochemistry and Physiology, 1975,50A, 99-104.

Cairns, R. B. Fighting and punishment from a devel-opmental perspective. In J. K. Cole & D. D. Jensen(Eds.), Nebraska Symposium on Motivation, (Vol.20). Lincoln: University of Nebraska Press, 1972.

Calhoun, J. B. The ecology and sociology of the Norwayrat. Bethesda, Md.: U.S. Department of Health,Education, and Welfare, 1962.

Conklin, P., & Heggeness, F. W. Maturation of tem-perature homeostasis in the rat. American Journalof Physiology, 1971,220, 333-336.

Cosnier, J. Le comportement du rat d'elevage. Un-published doctoral dissertation, University of Lyon,France, 1965.

Fairfield, J. Effects of cold on infant rats: Body tem-peratures, oxygen consumption, electrocardiograms.American Journal of Physiology, 1948, 155, 355-365.

Fowler, S. J., & Kellogg, C. Ontogeny of thermoregula-tory mechanisms in the rat. Journal of Comparativeand Physiological Psychology, 1975,89, 738-746.

Hahn, P., & Koldovsky, 0. Utilization of nutrientsduring postnatal development. Oxford: PergamonPress, 1966.

Hart, J. S. Rodents. In G. C. Whittow (Ed.), Compar-ative physiology of thermoregulation (Vol. 2). NewYork: Academic Press, 1971.

Howard, W. E. Relation between low temperature and

available food to survival of small rodents. Journalof Mammalogy, 1951,32, 300-312.

Hull, D. Thermoregulation in young mammals. In C. G.Whittow (Ed.), Comparative physiology of thermo-regulation (Vol. 2). New York: Academic Press,1973.

Jeddi, E. Confort du contact thermoregulation com-portementale. Physiology and Behavior, 1970, 5,1487-1493.

King, J. A., & Connon, H. Effects of social relationshipsupon mortality in C57BL/10 mice. PhysiologicalZoology, 1955,28, 233-239.

McCance, R. A. The maintenance of stability in thenewly born. Archives of Disease in Childhood, 1959,34, 459-470.

Pearson, 0. P. The oxygen consumption and bioener-getics of harvest mice. Physiological Zoology, 1960,33, 152-160.

Prychodko, W. Effect of aggregation of laboratory mice(Mus musculus) on food intake at different temper-atures. Ecology, 1958, 39, 500.

Randall, P. K., & Campbell, B. A. Ontogeny of behav-ioral arousal in rats: Effect of maternal and siblingpresence. Journal of Comparative and PhysiologicalPsychology, 1976, 90, 453-459.

Schmidt-Nielsen, K. Animal physiology. London:Cambridge University Press, 1975.

Sealander, J. A., Jr. The relationship of nest protectionand huddling to survival of Peromyscus at low tem-perature. Ecology, 1952, 33, 63-71.

Steineger, F. von. Beitrage zar soziologie and sonstigenbiologie der wanderratte. Zeitschrift fur Tierpsy-chologie, 1950, 7, 356-379.

Taylor, P. M. Oxygen consumption in new-born rats.Journal of Physiology, 1960,154, 153-168.

Welker, W. I. Factors influencing aggregation of neo-natal puppies. Journal of Comparative and Physio-logical Psychology, 1959,52, 376-380.

Whittow, G. C. (Ed.). Comparative physiology ofthermoregulation (Vol. 1). New York: AcademicPress, 1971.

Whittow, G. C. Evolution of thermoregulation. In G. C.Whittow (Ed.), Comparative physiology of thermo-regulation (Vol. 3). New York: Academic Press,1973.

Received March 7,1977 •

Documents

Huddling by rat pups: Group behavioral mechanisms of temperature regulation and energy conservation