High ambient temperature reduces rate of body-weight loss produced by wheel running

High ambient temperature reduces rate of body-weightloss produced by wheel running

Emilio GutierrezUniversidad de Santiago, Santiago de Compostela, Spain

Melissa T. BaysariUniversity of Sydney, Sydney, Australia

Olaia CarreraUniversidad de Santiago, Santiago de Compostela, Spain

Thomas J. Whitford and Robert A. BoakesUniversity of Sydney, Sydney, Australia

This study examined the effect of ambient temperature (AT) on the relationship between activity andweight loss. Compared with a neutral AT of 218C, high ATs of 27–298C produced a slower rate ofweight loss in rats given 1.5-hr food access and 22.5-hr running-wheel access in a standard activity-based anorexia (ABA) procedure (Experiments 1 and 2). The high AT did not affect food intake orwheel running in Experiment 1, but did reduce running in Experiment 2. Switching from neutral tohigh AT had only a transient effect on weight loss when wheel access was maintained (Experiment 2)but resulted in less weight loss when wheel access was prevented (Experiment 3). Giving rats only 3 hrof wheel access each day at a neutral AT also produced substantial weight loss, but less if for the rest ofeach day they were maintained at a high AT (Experiment 4).

When rats with restricted access to food are givenaccess to a running wheel, they run increasinglyeach day and can show progressive weight loss.The conditions that produce this self-starvationeffect (Routtenberg & Kuznesof, 1967) havebeen referred to as the activity-based anorexia(ABA) procedure (Epling, Pierce, & Stefan,1983). Comparisons across experiments suggestthat ambient temperature (AT) is important inmodulating the relationship between activity and

body weight, but the direct study of AT effectshas been neglected (Gutierrez, Vazquez, &Boakes, 2002). Understanding the role ofambient temperature in ABA experiments withrats may be important for treatment of anorexianervosa (Birmingham, Gutierrez, Jonat, &Beumont, 2004; Gutierrez et al., 2002).

Spontaneous running by rats with unrestrictedaccess to food increases as AT decreases(Stevenson & Rixon, 1957). This suggests that

Correspondence should be addressed to R. A. Boakes, School of Psychology, University of Sydney, NSW 2006, Australia.

Email: [email protected]

This research was partly supported by a grant (RAB) from the Australian Research Council and a scholarship (OC) from the Xunta

de Galicia, Spain. Experiment 1 was reported in an Honours thesis (MTB) submitted to the School of Psychology, University of

Sydney.

1196 # 2006 The Experimental Psychology Society

http://www.psypress.com/qjep DOI:10.1080/17470210500417688

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY

2006, 59 (7), 1196–1211

running might have a thermoregulatory function;that is, in some sense rats run to maintain theirbody temperature (e.g., Lambert, 1993).However, this cannot be the only factor involved,since Campbell and Lynch (1967) found thatrats run at ATs above their thermoneutral zone.This is the temperature range, from 27 to 308Cfor the rat, in which metabolic rate is minimal(Gordon, 1990). Rats maintained at 318C willrun even though this increases their hyperthermia(Campbell & Lynch, 1968). Furthermore, food-deprived rats increase running before any drop inbody temperature can be detected (Morrowet al., 1997).

Although this evidence shows that thermo-regulation cannot be the only factor controllingrunning in food-deprived rats, it leaves open thepossibility that AT is important for weight lossin an ABA experiment. In addition to between-experiment comparisons suggesting that higherATs produce slower weight loss (Gutierrez et al.,2002), some direct evidence was briefly reportedby Lambert (1993). She described a serendipitousfinding whereby an accidental increase in ATproduced a decrease in running speed. This wasfollowed by an unpublished ABA experiment inwhich AT was deliberately manipulated. Food-restricted rats with access to a running wheelwere tested at ATs of either 19.48C or 258C.Rats ran less and survived longer in the warmercondition (Lambert & Hanrahan, 1990).

Given the potential importance of AT, wesought to obtain more detailed evidence on itsrole in weight loss in the ABA procedure. Themain aim of the present study was to testwhether rate of weight loss is slower at a highAT (27–308C) than at a neutral AT (20–218C).This was achieved in Experiment 1 by comparingtwo groups of rats that were maintained at thesedifferent temperatures while running through astandard ABA procedure. A second aim was totest whether increasing AT once rats had lost con-siderable weight would slow down, or even reverse,this weight loss. This was achieved in Experiment 2where one group, neutral ! highAT, was startedat a neutral AT and then, when weights haddecreased to about 80% of their starting point,

the AT was raised to the higher level.Experiment 3 then sought to determine whethera high AT would aid recovery following removalfrom the ABA procedure. Here, two groups ofrats were maintained at either neutral or highATs while remaining on a restricted feeding sche-dule with no wheel access. The final aim was totest whether heating animals during periods ofinactivity would result in less running, moreeating, and consequently less weight loss. Thiswas achieved in Experiment 4 where two groupsof rats were maintained at either neutral orhigh ATs for 19.5 hr a day, whereas all received3 hr of wheel access and 1.5 hr of food at theneutral AT.

In addition to daily measurement of bodyweight and running, we also monitored foodintake. ABA experiments often find a closerelationship between changes in food intake andchanges in body weight (e.g., Dwyer & Boakes,1997). Consequently it was of interest to discoverwhether an effect of a high AT in decreasingthe rate of weight loss was accompanied byany effect of AT on food intake. Early studiesusing the ABA procedure tended to report onlybody weight and wheel running (e.g., Lambert,1993), and thus there are no previous data onthis issue.

EXPERIMENT 1

This experiment was designed to determine theimpact of AT on self-starvation. A standardABA procedure was used (Boakes & Dwyer,1997) while adopting a 2� 2 factorial designwith laboratory temperature set at either 218C(neutral) or at 298C (high). At each AT animalswere either allowed 22.5-hr wheel access (active)or remained in a stationary cage for the durationof the experiment (inactive). Because of the prac-tical difficulties of simultaneously maintainingsome rats at one temperature and others at a differ-ent temperature, experimental sessions for neutralgroups ended before those for highAT groupsbegan.

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2006, 59 (7) 1197

AMBIENT TEMPERATURE AND ACTIVITY-BASED ANOREXIA

Method

SubjectsThe subjects were naıve male albino Wistar rats.Whereas previously group housed, they weretransferred to single cages in the colony room 7days before their first experimental session.When the ABA procedure commenced (Day 0),the 16 animals allocated to the neutral conditionwere 53 days old, with a mean weight of 284.6 g,range 238.0–332.0 g, and the 16 allocated to thehighAT condition were 50 days old, with a meanweight of 258.9 g, range 229.0–300.0 g. Wateraccess was unrestricted in this experiment and inall subsequent experiments. Access to food (stan-dard rat chow in all experiments) was restrictedas described below.

ApparatusWhile in the colony room, rats were maintained ona 12-hr light–dark cycle, with lights on from 0800to 2000 hours. Illumination of the running-wheellaboratory was provided by natural lighting.

The laboratory contained 16 running wheels oftwo types. All were 10 cm wide with a circumfer-ence of 1.1 m. Eight, on an upper shelf, hadPerspex sides and floors made of metal rods.These were attached to acrylic cages,33� 21� 19 cm, with mesh lids in which waterbottles and food could be placed. The othereight, on a lower shelf, had solid metal walls andmesh floors. These were attached to mesh cages,25� 15� 12 cm, in which a water bottle couldbe inserted into a side wall. Food was placeddirectly into these side cages during feeding.Sliding doors controlled movement between thewheels and their side cages. Inactive subjectswere housed individually in acrylic cages,33� 21� 19 cm, with mesh lids in which awater bottle and food could be placed. All thecages were lined with wood shavings. Temperaturein the laboratory was recorded twice daily at 1000and 1500 hours. Over the 17 days in which rats inthe neutral condition were tested the meantemperature at 1000 hours was 20.68C, range20.0–23.08C, and at 1500 hours was 20.78C,range 20.0–23.08C. Over the 16 days in which

the highAT rats were tested the mean temperatureat 1000 hours was 29.18C, range 28.0–31.08C, andat 1500 hours was 29.48C, range 28.0–31.08C.

ProcedureWhile still receiving unrestricted access to food inthe colony room, rats within each AT conditionwere weighed daily for three days at 1000 hoursand then allocated to two weight-matched groupswhen transferred to the laboratory. During thefollowing 4-day preexposure period all rats contin-ued to receive unrestricted access to food. Thosein the active condition remained in the cagesattached to the running wheels for these four daysand were given access to the running wheels for2 hr each day. Rats in the inactive conditionremained in the individual acrylic cages.

The ABA procedure started (Day 0) with theremoval of food at 1500 hours and opening ofthe doors allowing access to the wheels in theactive group. From Day 1 onwards all rats weregiven access to food from 1330 to 1500 hours.For the active group the doors to the wheelswere closed during this period. Food intake wasmeasured by weighing the food at the beginningand end of every 1.5-hr feeding period. Ratswere also weighed daily at 1000 hours (as theywere on Day 0). Wheel turns were recordedevery 30 min by a computer. The period between1000 and 1330 hours, designated the food antici-patory period (FAP), was the period when thegreatest increase in running was expected tooccur (Dwyer & Boakes, 1997).

The experiment ended on Day 17 for theneutral groups and on Day 16 for the highATgroups, by which time most rats had reachedeither the removal criterion (75% of body weighton Day 0 for two days in succession) or the recov-ery criterion (weight on day n greater than weighton day n – 4). These criteria have been the stan-dard criteria for this kind of ABA experiment(Dwyer & Boakes, 1997; Routtenberg &Kuznesof, 1967).

Statistical analysisThe five dependent variables in all four of theseexperiments were: days to recovery, days to

1198 THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2006, 59 (7)

GUTIERREZ ET AL.

removal, body weight, food intake, and wheelturns. Differences in days to recovery andremoval were examined using nonparametricanalysis (Wilcoxon rank sum test). All othervariables were analysed using separate plannedcontrasts and trend analysis over the days untilthe first rat was removed. An alpha-rate of .05was used for all analyses.

Results

Body weightThe most important finding was that, while allactive animals at the neutral AT had to beremoved from the experiment, no active animals atthe high AT reached the removal criterion.Indeed, all animals housed at the high AT recov-ered, regardless of whether they were providedwith wheel access or not, but, as expected,highAT–active rats (n¼ 8) were slower to recover(12 median days) than highAT–inactive rats(n¼ 8; 8.5 median days), Mann-WhitneyU(8, 8)¼ 40, p, .01. A comparison between thetwo inactive conditions showed that the highAT–inactive group (n¼ 8; 8.5 median days) recoveredsooner than the neutral–inactive group (n¼ 3; 15median days), U(8, 3)¼ 36, p, .01. In the neutralAT condition active rats (n¼ 8; 8.5 median days)reached the removal criterion faster than the inac-tive animals (n¼ 3; 12 median days), U(8, 3)¼ 36,p, .01. Two neutral–inactive animals were stillhovering between the removal and the recovery cri-terion when the experiment ended.

A similar picture was shown by the changes inmean body weight over sessions, as shown in theupper panel of Figure 1. The weights of allgroups decreased over days, F(1, 28)¼ 306.40,but animals housed at the neutral AT showed amore rapid decline in weight than those at thehigh AT, F(1, 28)¼ 11.95. Active rats alsoshowed a more rapid decline than inactive rats,F(1, 28)¼ 22.13, and there was also a significantactivity by AT interaction, F(1, 28)¼ 5.69. Thisconfirmed that, as seen in Figure 1, the differencein body weight between active and inactive ratsbecame more pronounced at the neutral AT thanat the high AT.

Food intakeFood intakes over days are shown in the lowerpanel of Figure 1. As expected from previousABA experiments, active rats consumed less foodoverall than did inactive rats, F(1, 28)¼ 15.95,and also showed a less rapid increase in intakeover days, F(1, 28)¼ 28.48. However, AT hadno detectable impact on the amount of food con-sumed in either active, F(1, 28)¼ 1.44, or inactivegroups, F, 1.

Figure 1. Experiment 1. Top panel: Change over days in body

weight relative to weight on Day 0. “Active” denotes groups

given wheel access and “inactive” those without wheel access.

Bottom panel: Changes over days in food intake during the daily

1.5-hr feeding period.



Wheel turnsWheel turns were separated into those during the3.5-hr FAP, shown in the top panel of Figure 2,and those during the remaining 19 hr (1500hours the previous day to 1000 hours on the dayindicated on the graph), shown in the bottompanel. Running increased over days during boththe FAP and the overnight period, F(1,28)¼ 54.43; F(1, 28)¼ 35.92, respectively, butno group differences were found, Fs, 1.

Discussion

These results indicate that AT has a marked effecton body weight. All animals maintained at theneutral AT showed a greater reduction in weightthan did rats at the high AT. The difference inbody weight between active and inactive animalswas also greater at the low AT. This key finding—highlighted by the fact that no highAT–activeanimals had to be removed from the experimentwhereas all neutral–active rats were removed—indi-cates that high AT can prevent self-starvation. Thus,the present results provide a within-experiment con-firmation of the conclusion based on comparisonsacross different studies reached by Gutierrez et al.(2002) and on previous unpublished results(Lambert & Hanrahan, 1990).

An unanticipated finding was that AT had nodetectable effect on food intake, either on theactive or on the inactive groups. As for the activegroups, their reduced food intake relative to thatof the inactive rats reflects the consistent findingfrom the earliest ABA studies onwards thatrunning suppresses subsequent eating (e.g.,Routtenberg & Kuznesof, 1967).

A second unexpected finding was that AT didnot affect wheel running. Together with thesimilar lack of an effect of AT on food intake,this suggests that the rapid weight loss observedin active rats maintained at the neutral ATreflected their failure to increase food intake toprovide the calories needed to maintain bodytemperature. In this respect, the present resultsare not consistent with the thermoregulatoryexplanation of the self-starvation proposed mostrecently by Lambert (1993). This predicts thatthe tendency towards a lower body temperatureat the neutral AT should result in more runningin order to generate body heat, and this increasedactivity would in turn produce greater suppressionof food intake.

EXPERIMENT 2

One aim of this experiment was to provide afurther test for the possible effect of a high AT

Figure 2. Experiment 1. Top panel: Wheel turns per hour during

the 3.5-hr period prior to food access (food anticipatory period).

Bottom panel: Wheel turns per hour during remaining time

(19 hr) when the rats had access to the wheel.


GUTIERREZ ET AL.

on running and food intake. Thus, the initial stageof the experiment compared active groups main-tained at a neutral AT to those at a high AT, asin Experiment 1. However, an important differ-ence was that in Experiment 2 the rats were notgiven any preexposure to the wheels. The absenceof wheel preexposure produces lower runningrates in early sessions than when rats have alreadybecome familiar with the wheels prior to fooddeprivation (Boakes & Dwyer, 1997; Epling &Pierce, 1991; Pare, 1975), and slower runningmay be more sensitive to AT differences.

A second aim of this experiment was to testwhether increasing AT once rats had lost consider-able weight would slow down, or even reverse, thisweight loss. Such a reversal was briefly reported byLosada et al. (2001). Thus, in the present exper-iment one group, neutral ! highAT, was startedat a neutral AT of 218C and, when weights haddecreased to about 80% of their starting point,the AT was raised to 278C. A second group,highAT! neutral, was started at the high AT of278C and later switched to an AT of 218C. As inExperiment 1, experimental sessions for theneutral! highAT group ended before those forthe highAT! neutral group began.

Method

SubjectsTwo groups of naıve male albino Wistar rats(Concord Hospital, Sydney, NSW) were run con-secutively. They were transferred from group tosingle cages 7 days before the ABA procedurewas introduced (Day 0). On Day 0 the 12 rats inthe neutral! highAT group were 69 days old,with a mean weight of 371.0 g, range 324.0–402.0 g, and the 16 rats in the highAT!

neutral group were 70 days old, with a meanweight of 329.6 g, range 292.0–359.0 g. Accessto water and food was as described inExperiment 1. The ambient temperature of thecolony room was 218C (+18C).

ApparatusThe apparatus was the same as that used inExperiment 1. Temperature in the laboratory

was recorded twice daily at 1000 hours and 1500hours. For the neutral–highAT group the meantemperature over the first 6 days was 20.58C at1000 hours and 20.78C at 1500 hours; over theremaining days the mean temperature was27.48C at 1000 hours and 27.48C at 1500 hours.For the highAT–neutral group the mean temp-erature over the first 6 days was 27.38C at 1000hours and 27.38C at 1500 hours; over the remain-ing days it was 20.28C at 1000 hours and 20.88C at1500 hours.

ProcedureApart from the absence of preexposure to thewheels prior to food deprivation, the standardABA procedure was the same as that inExperiment 1, but no inactive controls wereincluded. Rats were initially kept in single cagesfor 3 days in the colony room where they wereweighed and handled daily. They were then trans-ferred to the laboratory to be kept in the side cagesfor a further 4 days. During this period the ratscontinued to have unrestricted access to food,but no access to the wheels.

The ABA procedure started at 1500 hours thenext day (Day 0) with removal of food andaccess to the wheels for the first time. From Day1 onwards access to food was given from 1330 to1500 hours, and rats were weighed daily at 1000hours. Other details were also as in Experiment 1.

For the neutral! highAT group the ambienttemperature was maintained at 208C from thetime that the rats first entered the laboratoryuntil the day (Day 6) that the first rat reachedthe body weight criterion of 80% or less of itsDay 0 weight for the second day in succession.This criterion was less severe than the standardremoval criterion—75% two days in succession—because animals were required to continue to asecond phase of the experiment; the ethical proto-col for these experiments insisted that, once thestandard removal criterion was reached, theanimals were to be removed from the experiment.

At 1500 hours on Day 6 the AT was increasedto 278C and was maintained at this level until theend of the experiment (1500 hours on Day 13).The procedure for the highAT! neutral group



was identical, except that the AT was 278C untilDay 6, when the temperature was decreased to208C for the rest of the experiment.

Results

Body weightBy the end of the experiment 3 rats had reached theremoval criterion, 2 from the neutral! highATand 1 from the highAT! neutral group. Norat reached the recovery criterion. Body weightsare shown in the top panel of Figure 3. OverDays 1–6 weight declined more rapidly inthe neutral! highAT group than in thehighAT! neutral group, F(1, 26)¼ 62.79.When ATs were reversed on Day 6, body weightdeclined more rapidly from Day 6 to Day 7 in thehighAT! neutral group, as confirmed by an inter-action between group and days, F(1, 26)¼ 148.84.Over Days 7–11 body weights remained higherin the highAT! neutral group, F(1, 26)¼ 7.13,and continued to decline, linear trend, F(1, 26)¼91.26, but the tendency for this decline to bemore rapid in the group switched to the neutraltemperature did not reach significance, in that theF-ratio for the interaction between group andtrend was 3.24, p¼ .08.

Food intakeOver Days 1–6 no overall difference in food intakewas detected between the groups, F(1, 26)¼ 2.58,but the interaction between group and trend wassignificant, F(1, 26)¼ 10.17, indicating that therate of increase of food intake was faster in theneutral! highAT group, as seen in the lowerpanel of Figure 3. The changes in AT on Day 6resulted in changes in intake, whereby the dropin AT for the highAT! neutral group wasaccompanied by a larger increase in food intakethan that in the neutral! highAT group, withan interaction between groups and days, F(1,26)¼ 20.57. Subsequent paired comparisonsshowed an increase in intake from Day 6 to Day7 by the highAT! neutral group, t(11)¼ 5.95,but no significant change in food intake fromDay 6 to Day 7 in the neutral! highAT group,t ¼ 1.57. As seen in Figure 3, from Day 7

onwards the group differences in intake weremaintained, F(1, 26)¼ 4.93, and intake continuedto increase, F(1, 26)¼ 102.44, but there was nolonger an interaction between group and trend,F(1, 26)¼ 2.88.

Figure 3. Experiment 2. Top panel: Change over days in body

weight relative to weight on Day 0. “Neutral!HighAT ” denotes

group with wheel access for which the AT was initially neutral

(218C) and then shifted to 278C, and “HighAT!Neutral ” the

group given these ATs in the opposite sequence. Bottom panel:

Changes over days in food intake during the daily 1.5-hr feeding

period. The vertical line between Days 6 and 7 indicates the

point at which the ambient temperature (AT) was increased for

the neutral!highAT group and decreased for the highAT!

neutral group.


GUTIERREZ ET AL.

Wheel turnsWheel turns were separated into those during the3.5-hr FAP, shown in the top panel of Figure 4,and those during the remaining 19 hr (1500 hoursthe previous day to 1000 hours on the day indicatedon the graph), shown in the bottom panel. It can beseen that during the FAP wheel turns increasedmore rapidly over Days 1–6 in the neutral!

highAT group, as shown by a main effect ofgroup, F(1, 26)¼ 11.65, an overall linear trend,F(1, 26)¼ 120.30, and most importantly a stronginteraction between group and trend, F(1, 26)¼26.02. Analysis of Days 6 and 7 revealed a maineffect of group, F(1, 26)¼ 5.50, and the importantinteraction between group and days, F(1,26)¼ 35.48. Separate analyses for each groupshowed that switching from neutral to high ATon Day 6 produced a decrease in running, t(11) =4.80; likewise, the change from high to neutralAT produced an increase in running, t(15)¼ 3.79.Analysis of Days 7–11 revealed only an overalltrend, F(1, 26)¼ 19.55. There was no longer anymain effect of group or interaction between groupand trend, Fs, 1.

As is typical, there was less wheel running perhour during the 19-hr postfeeding periods thanduring the 3.5-hr FAP prior to the feedingperiod. Nevertheless, the pattern of AT effectswas the same for postfeeding and for FAPrunning, as seen in Figure 4. Analysis of postfeed-ing wheel turns on Days 1–6 revealed that wheelturns increased more rapidly in theneutral! highAT group F(1, 26)¼ 7.22. OverDays 6 and 7 there was no main effect of group,F(1, 26)¼ 1.03, or of days, F, 1, but an inter-action between these factors, F(1, 26)¼ 61.81,whereby, as seen in Figure 4, running by theneutral! highAT group decreased when the ATwas raised, but running by the highAT! neutralgroup increased when the AT was reduced.Separate analyses for each group over Days 6–7showed that in the neutral! highAT group themean number of wheel turns decreased,t(5)¼ 3.87, and in the highAT! neutral groupwheel turns increased, t(7)¼ 8.34. Finally, overDays 7–11 an interaction between group andtrend, F(1, 26)¼ 4.31, indicated that wheel turnsincreased more rapidly in the neutral! highATgroup.

Discussion

Over Days 1–6 the higher AT produced a slowerloss of body weight and a slower increase inrunning. During this initial stage no overall

Figure 4. Experiment 2. Top panel: Wheel turns per hour during

the 3.5-hr period prior to food access (food anticipatory period).

Bottom panel: Wheel turns per hour during remaining time

(19 hr) when the rats had access to the wheel. As in Figure 3, the

vertical line between Days 6 and 7 indicates the point at which

the ambient temperature (AT) was increased for the neutral!

highAT group and decreased for the highAT!neutral group.



difference in food intake between the groups wasfound, but the rate of increase in food intake wassomewhat slower at the higher AT. Thus, oneaim of the experiment was achieved in thateffects of AT on food intake and wheel runningwere now found. In this case the slower rateof weight loss over Days 1–6 in thehighAT! neutral group is likely to reflect boththe lower activity level and the fewer caloriesneeded to maintain body temperature.

A further aim was to observe the effects ofreversing the ATs on Day 6. This produced agreater drop in weight from Day 6 to Day 7 inthe highAT! neutral than in the neutral!highAT group and a group difference in foodintake whereby this increased sharply when thetemperature dropped for the highAT! neutralgroup. The increase in temperature (neutral!highAT group) produced decreases in running,whereas the decrease in temperature(highAT! neutral group) produced the oppositeeffect. The AT-induced changes in activityoccurred both in the postfeeding and the FAPperiods.

Data from Days 7 to 11 showed that many ofthe immediate changes produced by AT reversalwere not sustained. Thus, body weight continuedto decline in the neutral! highAT group at arate that was only marginally different fromweight decline in the highAT! neutral groups.Similarly FAP wheel running continued at levelsand at rates of increase that were no longer detec-tably different between the two groups, while post-feeding running increased at a slightly higher ratein the higher AT (group neutral! highAT). Onthe other hand the higher food intake producedby AT reversal in the highAT! neutral groupwas sustained. This pattern suggests that,whereas food intake remains sensitive to thecurrent AT, this is not the case for activity orweight loss.

In conclusion, the first stage of this experimentextended the results of Experiment 1 by showingthat a high AT can reduce wheel running andalso that, after rats have lost considerable weightwhen maintained at a neutral AT, switching to ahigher AT can produce a reduction in wheel

running and slow down the rate of weight loss,at least temporarily. However, it seems that bythis stage running had acquired for these rats acompulsive character so that it overrode thehomeostatic processes that normally regulatebody temperature, weight, and calorie intake. Inthis respect the present finding contrasts with thebrief report that increasing the AT from 208C to278C produced recovery of body weight (Losadaet al., 2001).

EXPERIMENT 3

Once rats have lost 25% or more of their weight asa result of exposure to the ABA procedure, onlytwo manipulations have proved effective inprolonging survival (Gutierrez et al., 2002):either the combination of denying access tothe wheel and giving unrestricted access to food(Boakes, Mills, & Single, 1999; Pare, 1976) ormaintaining wheel access and restricted feeding,but increasing AT well over the thermoneutralzone (378C; Morrow et al., 1997). Given that inthe previous experiment increasing the AT failedto stop further weight loss when wheel accessremained available, the aim of the present exper-iment was to determine whether an increase inAT might have an impact if at the same timeaccess to the wheels was now prevented.

In this experiment rats were first given pre-exposure to the wheels and were then exposed tothe ABA procedure used in Experiments 1 and 2.For all rats the AT was initially at a neutral level.When a rat reached the removal criterion, itsaccess to the running wheel was prevented, butthe 1.5-hr restricted feeding schedule was main-tained. At the same time for half the rats the ATwas increased to a high level, whereas for theremainder the neutral AT was maintained. Asdescribed below, the use of infrared lampsmounted above each side cage enabled us tocontrol the effective AT for each rat indepen-dently. This change in equipment was introducedto overcome the limitation imposed on theprevious two experiments, such that on a given


GUTIERREZ ET AL.

day the effective AT was the same for all rats—namely, the AT set for the laboratory as a whole.

A further change introduced in the presentexperiment was to use female hooded rats. Thiswas because albino males of the kind used inExperiments 1 and 2—and in most studies ofABA—were not available at the time. Thus, thepresent subjects differed in potentially importantways from the rats used in Experiments 1 and 2.While this precluded meaningful between-experiment comparisons, it contributed towardstesting the generality of the AT effects examinedin this project.

Method

SubjectsA total of 16 female hooded Wistar rats were 91days old, with a mean body weight of 188.3 g,range 169.5–210.0 g, on Day 0.

ApparatusThe set of running wheels was that used inExperiments 1 and 2, and the only modificationwas to mount 50-W infrared lamps above eachside cage. When the AT for the laboratory wasset at 218C, these lamps increased the air tempera-ture within a side cage to a mean of 29.58C, range28–308C. Over the whole experiment the meantemperatures in the laboratory were 20.58C,range 19–228C, at 1000 hours, and 20.78C,range 19–238C, at 1500 hours.

ProcedureRats were initially handled and weighed daily for 3days while in the colony room, which was main-tained at 228C. They were then transferred tothe laboratory where the AT was set at 208C.All animals were then given the preexposure treat-ment, as in Experiment 1, followed (Day 0) by thestart of the standard ABA procedure: food accessrestricted to 1.5 hr daily and otherwise access tothe wheels. This phase continued for each ratuntil it reached an initial criterion of 80% of Day0 body weight. On reaching the 80% criterion, arat was prevented from reentering its running

wheel, but remained on the restricted-feedingschedule.

Once an animal reached criterion, it wasassigned to one of two groups, highAT orneutral, matched on days to criterion by allocatingthe first rat to the highAT condition (its infraredlamp was switched on for the remainder of theexperiment), the second to the neutral condition(no infrared lamp), the third to the neutral con-dition, the fourth to the highAT condition, andso on. These conditions were maintained untilrats reached either the recovery criterion or theremoval criterion (75% of Day 0 body weight for2 days in succession).

Results

In the initial ABA stage all rats lost weight rapidly,as expected since all were maintained at the neutralAT throughout this stage. Body weight, foodintake, and activity were analysed in this firstphase prior to the removal of first rat from therunning wheel (Day 3). Although body weightdecreased over days, F(1, 14)¼ 602.84, this trenddid not interact with the to-be-assigned condition,F, 1. Similarly, no group differences were foundin food intake, F(1, 14)¼ 1.90, or in wheel turnsduring this first phase, F, 1. The effectivenessof the matching procedure was indicated by thesame median number of days to the initial 80%criterion, 5, for both groups.

The important results were that in the followingphase half the rats in the highAT condition recov-ered (n¼ 4; 8 median days), whereas none of theneutral rats did so, and that neutral rats (n¼ 8; 3median days) reached the removal criterion signifi-cantly faster than the highAT rats (n¼ 4; 4.5median days), Mann-Whitney U(8, 4) ¼ 43,p, .01.

Mean body weights were calculated for each ofthe 3 days prior to removal of the first neutral rat,as shown in Figure 5. There was an overall differ-ence in body weight between neutral and highATanimals over these three days, F(1, 14)¼ 16.95,an overall linear downward trend, F(1, 14)¼77.07, and a group by trend interaction, indica-ting that body weight in the highAT group



declined more slowly than that in the neutral group,F(1, 14)¼ 11.08. During this second stage neutralrats also consumed more food on average (M ¼

8.09 g, SEM ¼ 0.18 g) than did highAT rats (M¼ 6.91 g, SEM ¼ 0.30 g), F(1, 14)¼ 10.16.

Discussion

When wheel access was removed, raising the ATdecreased the rate of weight loss and evenallowed weight recovery in half the animals. Thisfinding is consistent with that of Morrow et al.(1997) using somewhat different conditions,including switching to a much higher AT(378C). These authors also reported that, oncemarked weight loss had occurred, denial of wheelaccess was not sufficient to keep the rats alive.Even though the initial criterion in the presentexperiment, 80% of Day 0 weight, was much lesssevere than that used by Morrow et al. (1997), itwas notable that all 8 rats in the neutral group con-tinued to lose weight even after wheel access wasdenied.

EXPERIMENT 4

A variation on the standard ABA procedure is onein which rats are given wheel access for only a fewhours daily. Using this approach Boakes andDwyer (1997) found that male albino ratshoused in groups of four when they were not inthe wheel showed less weight loss than ratshoused singly. They speculated that the social iso-lation of single-housed rats might produce chronicstress that could accentuate the effects of activityon body weight (cf. Brown & Grunberg, 1995;Richard & Rivest, 1989). A similar result hasbeen reported for both male and female hoodedrats (Boakes et al., 1999; Exp. 2).

The present experiments suggest an alternativeexplanation for the housing effect to the onesuggested by Boakes and Dwyer (1997). Giventhat rats tend to huddle together much of thetime when group housed, it is likely that thisreduces the heat loss that a single-housed ratwould suffer when the AT is below its thermoneu-tral level. It follows that maintaining rats at a highAT when they are not in the wheel could have thesame effect as housing them in groups. Thepresent experiment was undertaken to test thispossibility. In the active condition rats weregiven 3-hr wheel access followed by 1.5-hr foodaccess each day at a neutral temperature, whileinactive rats spent this time in the side cages atthe same temperature. The main new factor inthis 2� 2 design was the AT during the restof the day: For the highAT–active andhighAT–inactive groups this was increased byswitching on the infrared lamps described forExperiment 3 so as to raise the effective AT toaround 298C, while for the neutral–active andneutral–inactive groups the temperature wasnever raised above the neutral AT of the laboratory.

Method

SubjectsAs in Experiment 3, hooded female Wistar ratswere used. The experiment was run in two rep-lications, each containing 16 rats. In the first

Figure 5. Experiment 3. Changes in body weight relative to Day 0

weight for the first 3 days after each rat reached an initial criterion

of 80% following exposure to a standard ABA procedure. At this

point rats were denied further access to the wheel and were

allocated to groups maintained at either a neutral or a highAT.


GUTIERREZ ET AL.

replication the rats were 76 days old on Day 0,with a mean weight of 176.8 g, range 156.5–189.0 g, and in the second replication the ratswere 82 days old, with a mean weight of 180.7 g,range 165.6–195.5 g, on Day 0.

ApparatusThe running wheels and arrangement of infraredlamps over the side cages were the same as thoseused in Experiment 3, with the same temperatureswithin the side cages when the lamps were on—namely, a mean of 29.58C, range 28–318C. Themean ATs in the laboratory over the duration ofthe experiment were 20.68C, range 19–238C, at1000 hours and 20.88C, range 20–228C, at 1500hours.

ProcedureWithin each replication rats were initially weigheddaily for 3 days while still group housed in thecolony room, and then 4 rats were allocated toeach of the four conditions on a weight-matchedbasis. At 1430 hours on Day 0 all rats were trans-ferred to the laboratory and were placed in the sidecages: No food was available, and the infraredlamps were switched on for animals allocated tothe highAT condition. From Day 1 onwards, allrats were weighed at 0940 hours, and from 1000 to1300 hours active rats were given access to therunning wheel, while inactive rats remained in theside cages. All rats were fed daily in the side cagesfrom 1300 to 1430 hours. Rats were maintained ateither the high or the neutral AT, except duringthe running and feeding periods—total of 4.5 hr—when all rats were at the neutral AT (all infraredlamps were switched off). These conditions weremaintained until rats reached either the recovery cri-terion or the removal criterion (75% of Day 0 bodyweight two days in succession).

Results

Body weightAll 16 neutral rats reached the removal criterion,while only 2 rats in the highAT condition wereremoved from the experiment (in 11 mediandays). Of the remaining rats in this condition 11

reached the recovery criterion, while 3(highAT–active) remained in an intermediatestate, neither gaining nor losing weight. In termsof days to the removal criterion rats in thehighAT condition remained in the experimentlonger than rats in the neutral condition, Mann-Whitney U(16, 16)¼ 138.5, p , .01. There wasno difference between highAT–active (n¼ 5; 13median days) and highAT–inactive (n¼ 6; 12median days) groups in days to reach the recoverycriterion, U(5, 6) ¼ 30.5, and the differencebetween neutral–active (n¼ 8; 7.5 median days)and neutral– inactive (n¼ 8; 9.5 median days)groups in days to reach the removal criterion onlyapproached significance, U(8, 8)¼ 50, p¼ .053.

Mean body weights relative to Day 0 weight areshown in Figure 6 for the 5 days prior to removalof the first rat. It may be seen that weights declinedmore rapidly in the neutral condition and morerapidly in active than in inactive groups. Thiswas confirmed by analysis showing a main effectof high versus neutral AT, F(1, 28)¼ 14.59, anda linear trend over days, F(1, 28)¼ 1,372.91,which differed across groups. Active rats showed

Figure 6. Experiment 4. Change over days in body weight relative

to weight on Day 0. As in Figures 1 and 3, “active” denotes groups

given wheel access and “inactive” those without wheel access.

However, in this experiment “neutral” and “highAT’ refer to ATs

during the period when the rat had access neither to the wheel nor

to food.



a more rapid decline in body weight than inactiveanimals, F(1, 28)¼ 11.13, and neutral animals alsoshowed a more rapid decline in body weight thandid highAT animals, F(1, 28)¼ 35.13, but therewas no significant interaction between thesefactors, F, 1. Thus, there was no evidence thatthe high between-session temperature reducedthe difference in rate of weight loss betweenactive and inactive groups.

Food intakeInactive animals consumed more food overall thandid active animals, F(1, 28)¼ 5.34, thus confirm-ing that even relatively brief periods ofrunning—3 hr in this case—can reduce subsequentfood intake. There was also an overall increase infood intake over days, as shown in the upperpanel of Figure 7, F(1, 28)¼ 372.47. A significantthree-way interaction between the groups andlinear trend indicated that activity had a greaterimpact on the rate of increase of food intake atneutral AT than at high AT, F(1, 28)¼ 6.11.

Wheel turnsAs shown in the lower panel of Figure 7, wheelrunning increased over days, F(1, 14)¼ 89.79,and neutral–active animals ran more overall thandid highAT–active animals, F(1, 14)¼ 66.28,but no group by trend interaction was detected,F(1, 14)¼ 3.07.

Discussion

The results show clearly that keeping rats at a highAT during that part of the day when they are notrunning can attenuate the effects of running onbody weight loss. This finding is comparable tothat reported when rats were housed communallywhen not placed in running wheels (Boakes &Dwyer, 1997). Consequently a parsimoniousexplanation for the two sets of findings is thatboth a heat lamp and other rats can raise the effec-tive AT for a rat and thus reduce the utilization ofcalories for maintaining body temperature.

GENERAL DISCUSSION

The principal findings from these four exper-iments were as follows. A high AT reduced therate of weight loss produced by the ABA pro-cedure in Experiments 1 and 2. In addition,Experiment 2 also found less running at the highAT. When the AT was switched from neutral tohigh, this had only a transient effect on weightloss and running in Experiment 2. However, when

Figure 7. Experiment 4. Top panel: Changes over days in food

intake during the daily 1.5-hr feeding period. Bottom panel:

Wheel turns per hour during the 3-hr daily period when the rats

had access to the wheel. “Active” denotes groups given wheel access

and “inactive” those without wheel access.


GUTIERREZ ET AL.

in Experiment 3 the increase in AT occurredwhen wheel access was denied after the rats’weights had declined to 80% of their initial level,raising the AT retarded mean weight loss and pro-duced recovery of body weight in half the subjects.Finally, Experiment 4 found that maintaining ratsat a high AT during the 19.5 hr per day when theywere neither running nor eating also retardedweight loss. The results are of interest for twomain reasons. One is their contribution towardsunderstanding the effects of ambient temperatureon the relationships between activity, food intake,and body weight. The second is their possible rel-evance to the treatment of anorexia nervosa (AN).

Loss of body weight produces increasedrunning, and under ABA conditions this canlead to steadily decreasing weight for two mainreasons. First, recent activity reduces food intakeand reduces the rat’s ability to adjust to therestricted feeding schedule (Dwyer & Boakes,1997; Kanarek & Collier, 1983). Raising the ATreduces rate of weight loss, as seen in Experiments1 and 2, and, since there was little effect on foodintake, this must have been partly because therewas less utilization of calorie reserves to defendcore temperature. Thus, in Experiment 1 thehigh AT seems to have given the group exposedto the ABA procedure (highAT–active group)the extra time needed to adapt to the restrictedfeeding schedule, in the sense of increasing theirdaily food intake, and thus to recover weightdespite continuing to run at a high level.

The second way in which AT may affect bodyweight loss is indicated by Experiment 2. Evenwhen food access is not restricted, rats—as wellas most other animals—normally spend consider-able time running when given access to arunning wheel (Sherwin, 1998), and their weightremains below that of inactive controls so long aswheel access is continued (Afonso & Eikelboom,2003). Furthermore, although wheel access isinitially accompanied by suppression of foodintake for about 8 days, such rats eventually eatmore than control animals without access to awheel (Lattanzio & Eikelboom, 2003). This andother evidence (e.g., Lett, Grant, Smith, & Koh,2001) indicate that, whether or not a rat is

adapting to a new feeding schedule, its level ofactivity can have a direct impact on body weight.In Experiment 2 the group given a high AT atthe start ran less, and this must have contributedto their slower loss of weight.

As for the effects of AT on rats without wheelaccess in the present experiments, these can berelated to some important data on the effect offood deprivation on rats’ preferred AT. Suchtemperature preferences have been measuredusing a long chamber that is cold at one end andhot at the other, with a continuous thermal gradi-ent throughout. When such equipment withextremes of 15 and 348C was used to assesstemperature preferences as a function of fooddeprivation, Sakurada et al. (2000) found thatnonstarved controls preferred 21–228C—the“neutral” AT used in the present experiments—whereas after 48 hr without food rats preferredabout 288C in the light period, corresponding tothe “high” AT used here. These authors concludedthat seeking out a higher temperature, particularlyduring the light cycle, functions as a behaviouraladaptation that “may contribute to reducing meta-bolic heat production and thus save energy underfood deficit conditions” (Sakurada et al., 2000,p. 423). This is consistent with a finding fromthe same group that food deprivation increasescold-escape behaviour (Yoda et al., 2000).

The above findings may illuminate the datafrom inactive rats in the present experiments inthe following way. Body weight can normallyrecover from the initial impact of restricting foodaccess to 1.5 hr per day under neutral AT con-ditions, as shown by the neutral–inactive groupin Experiment 1 (see Figure 1, top panel). Onthe other hand, once body weight has droppedbelow 80% recovery may no longer take place atthis AT, as shown by the neutral group inExperiment 3, which was denied access to thewheel once this amount of weight loss hadoccurred. The obvious explanation both for theseresults and for the data reported by Sakuradaet al. (2000) and Yoda et al. (2000) is that bodyweight loss involves reduced thermal insulation.Hence, increasing the AT well above the neutrallevel can slow down or reverse further weight



loss, whether maintained throughout the day, as inExperiment 3, or for the 19.5 hr per day when therats were not either running or eating, as inExperiment 4. Incidentally, the finding that pre-ferred AT increases with food deprivation(Sakurada et al., 2000) suggests that, althoughthe AT commonly maintained in rat colonyrooms, 20–228C, may be entirely appropriate fornondeprived rats, a somewhat higher AT may beoptimal for food-deprived rats unless they aregroup housed.

In this context it may be noted that the inactiverats in Experiments 1 and 4 lost weight morerapidly than expected on the basis of previousexperiments. Thus, of the 8 albino males in theneutral–inactive group in Experiment 1, 5 failedto reach the recovery criterion, whereas all 16rats of almost the same age and weight recoveredwhen given very comparable conditions in previousexperiments (Boakes & Juraskova, 2001, Exps. 1and 2). Similarly, all 8 hooded females in theneutral–inactive group of the present Experiment4 reached the removal criterion, whereas out of acomparable group of 8 single-housed females ina previous experiment only 1 failed to reach therecovery criterion (Boakes et al., 1999, Exp. 2).The only factor that appears to account for thisdiscrepancy is that of ambient temperature; theprevious experiments were carried out in ATsthat ranged from 22–248C. It is therefore possiblethat relatively small differences in temperatures inthis range can have a large effect on how individu-ally housed rats react to being placed on a 90-mindaytime feeding schedule.

Following the pioneering work of Routtenbergand Kuznesof (1967), it has been known that, oncean ABA procedure has produced a weight loss ofmore than 25%, a rat will die unless these con-ditions are discontinued. One aim of Experiment 2was to test the suggestion (Lambert, 1993;Losada et al., 2001) that switching to a higherAT can reduce wheel running after rats have lostconsiderable weight following exposure to theABA procedure at a neutral AT. Only a transienteffect of this kind was detected, and, instead, underpresent conditions AT appears to control the rateat which running increases rather than activity

level per se (see top panel of Figure 2). Itremains possible, of course, that with a differentset of conditions—including control of individualAT levels as in Experiments 3 and 4—a more sus-tained reduction in running might be obtained byswitching from neutral to a high AT.

With regard to clinical implications, thepresent results are important for nonpharmacolo-gical management of human AN patients. Theuse of ABA-induced self-starvation in rats as ananimal model of AN is in conflict with the histori-cally influential view that excessive exercise in ANis, first and foremost, purposeful behaviour bypatients desiring to lose weight. This contrastswith a psychobiological perspective from whichincreased activity is seen as a core symptom(Davis, 1997; Gutierrez et al., 2002). Hyperactiv-ity and excessive exercise present a major problemin the management of AN patients. Holtkamp,Hebebrand, and Herpertz-Dahlmann (2004)recently concluded that anxiety and food restric-tion contribute synergistically to increased levelsof physical activity in the acute phase of AN.When heat treatment is used with AN patients,this seems to dampen their autonomic hyperactiv-ity (Birmingham et al., 2004). The effects of ATon body weight obtained in the present exper-iments suggest that keeping AN patients warmmay slow down body weight loss, or even allowsome weight gain, in spite of a patient’s restrictedeating.

Original manuscript received 14 April 2005

Accepted revision received 6 September 2005

PrEview proof published online 15 February 2006

REFERENCES

Afonso, V. M., & Eikelboom, R. (2003). Relationshipbetween wheel running, feeding, drinking, andbody weight in male rats. Physiology & Behavior,80, 19–26.

Birmingham, C. L., Gutierrez, E., Jonat, L., &Beumont, P. (2004). Randomized controlled trialof warming in anorexia nervosa. International

Journal of Eating Disorders, 35, 234–238.


GUTIERREZ ET AL.

Boakes, R. A., & Dwyer, D. M. (1997). Weight loss inrats produced by running: Effects of prior experienceand individual housing. Quarterly Journal of

Experimental Psychology, 50B, 129–148.Boakes, R. A., & Juraskova, I. (2001). The role of drink-

ing in the suppression of food intake by recentactivity. Behavioral Neuroscience, 115, 718–730.

Boakes, R. A., Mills, K. J., & Single, J. P. (1999). Sexdifferences in the relationship between activity andweight loss in the rat. Behavioral Neuroscience, 113,1–10.

Brown, K. J., & Grunberg, N. E. (1995). Effects ofhousing on male and female rats: Crowding stressesmales but calms females. Physiology & Behavior, 58,1085–1089.

Campbell, B. A., & Lynch, G. S. (1967). Activity andthermoregulation during food deprivation in therat. Physiology & Behavior, 2, 311–313.

Campbell, B. A., & Lynch, G. S. (1968). Influence ofhunger and thirst on the relationship between spon-taneous activity and body temperature. Journal of

Comparative and Physiological Psychology, 65, 492–498.Davis, C. (1997). Eating disorders and hyperactivity: A

psychobiological perspective. Canadian Journal of

Psychiatry, 42, 168–175.Dwyer, D. M., & Boakes, R. A. (1997). Activity-based

anorexia in rats as a failure to adapt to a feedingschedule. Behavioral Neuroscience, 111, 195–205.

Epling, W. F., & Pierce, W. D. (1991). Solving the anor-

exia puzzle. A scientific approach. Toronto, Canada:Hogrefe & Huber.

Epling, W. F., Pierce, W. D., & Stefan, L. A. (1983).A theory of activity-based anorexia. International

Journal of Eating Disorders, 3, 27–46.Gordon, C. J. (1990). Thermal biology of the laboratory

rat. Physiology & Behavior, 47, 963–991.Gutierrez, E., Vazquez, R., & Boakes, R. A. (2002).

Activity-based anorexia: Ambient temperature hasbeen a neglected factor. Psychonomic Bulletin &

Review, 9, 239–249.Holtkamp, C., Hebebrand, J., & Herpertz-Dahlmann,

B. M. (2004). The contribution of anxiety and foodrestriction to physical activity levels in acute anorexianervosa. International Journal of Eating Disorders, 36,163–171.

Kanarek, R. B., & Collier, G. H. (1983). Self-starvation:A problem of overriding the satiety signal? Physiology

& Behavior, 30, 307–311.Lambert, K. G. (1993). The activity-stress paradigm:

Possible mechanisms and applications. Journal of

General Psychology, 120, 21–32.

Lambert, K. G., & Hanrahan, L. (1990, April). The

effect of ambient temperature on the activity-stress

ulcer paradigm. Paper presented at the meeting ofthe Southern Society for Philosophy andPsychology, Louisville, KY, USA.

Lattanzio, S. B., & Eikelboom, R. (2003). Wheel accessduration in rats: I. Effects on feeding and running.Behavioral Neuroscience, 117, 496–504.

Lett, B. T., Grant, V. L., Smith, J. F., & Koh, M. T.(2001). Preadaptation to the feeding schedule doesnot eliminate activity-based anorexia in rats.Quarterly Journal of Experimental Psychology, 54B,193–199.

Losada, M. C., Freire, M., Pardinas, M. C.,Vazquez, M. J., Pereiro, M. D., Nunez, M. J., et al.(2001, April). Effect of temperature modifications on

the development and reversibility of the activity-stress

paradigm in rats: Relationship to anorexia nervosa

therapy? Paper presented at the InternationalBehavioral Neuroscience Society, Cancun, Mexico.

Morrow, N. S., Schall, M., Grijalva, C. V.,Geiselman, P. J., Garrick, T., Nuccion, S., et al.(1997). Body temperature and wheel runningpredict survival times in rats exposed to activity-stress. Physiology & Behavior, 62, 815–825.

Pare, W. P. (1975). The influence of food consumptionand running activity on the activity-stress ulcer in therat. American Journal of Digestive Diseases, 20, 262–273.

Richard, D., & Rivest, S. (1989). The role of exercise inthermogenesis and energy balance. Canadian Journal

of Physiology and Pharmacology, 67, 402–409.Routtenberg, A., & Kuznesof, A. (1967). Self-starvation

of rats living in activity wheels on a restricted feedingschedule. Journal of Comparative and Physiological

Psychology, 64, 414–421.Sakurada, S., Shido, O., Sugimoto, N., Hiratsuka, Y.,

Yoda, T., & Kanosue, K. (2000). Autonomic andbehavioural thermoregulation in starved rats.Journal of Physiology, 526, 417–424.

Sherwin, C. M. (1998). Voluntary wheel running: A reviewand novel interpretation. Animal Behavior, 56, 11–27.

Stevenson, J. A., & Rixon, R. (1957). Environmentaltemperature and deprivation of food and water onthe spontaneous activity of rats. Yale Journal of

Biological Medicine, 29, 575–584.Yoda, T., Crawshaw, L. I., Yoshida, K., Su, L., Ozono, K.,

Shido, O., et al. (2000). Effects of food deprivationon daily changes in body temperature and behavioralthermoregulation in rats. American Journal of

Physiology: Regulatory Integrative Comparative

Physiology, 278, R134–R139.



High ambient temperature reduces rate of body-weight loss produced by wheel running

Documents