Effects of testosterone upon feeding in male mice

Anim . Behav., 1978,26,945-952

EFFECTS OF TESTOSTERONE UPON FEEDING IN MALE MICE BY STEWART PETERSEN

Department ofPhysiology, University of Leicester, University Road, Leicester, LEI 7RH Abstract. Food intake and pattern of feeding were recorded in intact and castrate male mice . Intact adults are larger and eat more than pre-pubertal castrates, due to rapid growth at puberty . Castration in adulthood has little effect upon food intake, but feeding becomes concentrated into large, infrequent meals . The change in meal size appears primary and is reversed by testosterone. During rapid growth at puberty, both meal size and frequency are high . It is suggested that testosterone affects feeding in two ways . Rapid weight gain at puberty leads to high meal size and frequency, but in adulthood meal size is depressed . Meal size in adults is further depressed by stimuli from females, and it is argued in this case that testosterone affects a general mechanism for altering feeding when other activities have a high priority . The gonadal hormones have been observed to have considerable influence over the food intake and feeding behaviour of laboratory rodents . So far, much attention has been concentrated upon the role of oestrogens, which are thought to produce a reversible depression of body weight `set point', (Wade 1972), but relatively little attention has been devoted to the male hormones such as testosterone . In this paper, the effects of testosterone upon the feeding of male mice will be examined . There are, in principle, two ways in which gonadal hormones might be expected to affect feeding. Firstly, they are intimately involved in the production and maintenance of sexual dimorphisms in body size, and secondly, they produce a number of other behavioural effects which may lead directly or indirectly to changes in feeding behaviour. The development of sexual dimorphisms has been examined in detail for the rat by Bell & Zucker (1971) . These authors gonadectomized male and female animals at birth, and found that these neonatal castrates achieved adult body weights intermediate between those of intact male and female controls . Influences of both the ovaries and testes are therefore important for the normal sexual dimorphism . Administration of testosterone to adult neonatal castrates leads to a rise in body weight towards that of the intact male, and this is accompanied, not unexpectedly, by increased food intake . The major part of the difference in body weight between neonatal castrates and control males appears in a relatively short period after puberty, when male animals grow very rapidly, whereas after this stage the intact animal maintains a rate of growth slightly in excess of that of his castrate counterpart . Furthermore, this growth

is irreversible, as castration of adult males leads to only a very slight fall in body weight . (Kakolewski et al . 1968) . Testosterone appears, therefore, to produce a single, rapid, irreversible promotion of body weight (and food intake) when it is present at puberty, and thereafter promotes a rate of weight gain only slightly in excess of the castrate pattern. The evidence to date suggests that the same is true for laboratory mice . Males do gain weight rapidly after puberty, but not much faster than females thereafter, and castration of adults leads to only a slight fall in body weight (Wright & Turner 1973) . In addition to being larger than their neonatally castrated counterparts, intact males show a number of behavioural differences . They are more likely to indulge in aggressive and sexual behaviour for example . It might be expected that these behavioural propensities would influence the activities of feeding. Such a phenomenon is indeed observed in an extreme form in certain species such as the red deer, where the stag eats practically nothing during the annual rut . The present author has suggested elsewhere (Petersen 1976) that modulations of feeding over the oestrous cycle of the female mouse may be divided into two similar classes on the basis of examination of the detailed temporal pattern of feeding . Changes when the female is on heat appear to influence mainly meal size, whereas after ovariectomy and towards the end of the cycle, meal frequency is the only variable altered . Accordingly. in this study the temporal pattern of feeding of male mice, and their castrate counterparts was observed under a variety of circumstances . 945

9 46

ANIMAL BEHAVIOUR, 26, 3

Methods Subjects in all experiments were C57 mice of stock originating from the laboratory animals centre, University of Edinburgh . Prior to the study, males were caged individually in a temperature-controlled room (23 + 2 C) on a 10-hour dark, 14-hour light, cycle . The experiments performed may be divided into a number of categories . (1) Effects of Testosterone Upon Food Intake and Body Weight In all these experiments food intake was measured by regularly weighing a feeder insert in the animal's home cage . The cages were of a standard type, and each contained a single mouse. They had a formed polypropylene base, and a galvanized top with depressions for food and water. Perspex feeders were constructed to fit into the food depression so that the animals could feed in the normal way from a receptacle that could be removed and weighed . The feeder was weighed once each day at the same time . Water was available ad libitum . Spillage was measured on a proportion of days, but, probably because of the difficulty of feeding from the feeders, never exceeded 0 . 1 g/day . (1a) Effects of castration before puberty . Two groups of eight males were castrated under Equithesin anaesthesia at 21 days of age . A third group of eight were given a sham operation at the same time . The food intake and body weight of each animal in one of the castrate and the control groups was followed for 5 weeks after the operation . The remaining eight castrated animals were left for 4 weeks, and then treated with seven consecutive daily doses of 20 gg testosterone propionate in oil by intramuscular injection . Their food intake was measured for I week before, during, and I week after the injections . (ib) Effects of castration in adulthood. Twelve males were castrated, again under Equithesin anaesthesia, at 96 days of age, and their food intake and body weight monitored for 3 weeks following the operation . The intake and body weight of a group of 12 intact control animals was also tested . Half of the castrated animals then received seven daily injections of 20 pg testosterone propionate in oil (intramuscular), and the other half were given injections of oil . Food intake and body weight were recorded throughout .

(2) The Effects of Testosterone Upon the Temporal Pattern of Feeding The temporal pattern of feeding was recorded by a method described previously (Petersen 1976), with no modification . Basically, the animal must push open a light Perspex door in a feeder suspended from the roof of its cage . Movements of the door are recorded and sensed electrically, and the resulting data are made up of collections of successive bout lengths and interbout intervals, and are analysed into a number of summary variables, such as meal size and intermeal interval, whose complete definition is given by Petersen (1976) . In all cases animals fed from the feeder recorders for at least 1 week before any recording began in order to ensure that they were well adapted to the situation . (2a) Effects of castration in adulthood on feeding pattern. The feeding behaviour of eight adult males was recorded continuously for 3 days after the usual period of acclimatization . They were then castrated under Equithesin anaesthesia, allowed 2 weeks to recover, and feeding recorded again for 3 days . Variables of feeding such as mean meal size, mean meal frequency, mean bout duration, etc . were calculated for each animal, and compared with the situation before castration using paired Student's t-tests . (2b) Effects of injected testosterone upon the feeding of males castrated as adults . Eight adult males were castrated, and after allowing 2 weeks for them to recover their feeding patterns, were recorded as above for 3 days . They were then randomly divided into four groups of two, and each pair received four different doses of hormone, in each case for four consecutive days by intramuscular injection, separated by 10 days' rest . The doses were 0 . 1 ml oil, 5, 15 and 25 pg testosterone propionate . The doses were given in counterbalanced order to the different groups . Feeding pattern data were recorded for all days of injection, and the mean pattern variables calculated for each animal on each dose, so that in total feeding pattern data were available for eight animals on each of the doses employed. The data were analysed using Friedman's test . (2c) The effects of testosterone injection upon adult males castrated before puberty . Six male mice were castrated at 21 days of age, and allowed to develop for 4 weeks . Their feeding was recorded for 3 days, and the following week



PETERSEN : TESTOSTERONE AND FEEDING IN MALE MICE

all six received four consecutive daily intramuscular injections of 20 µg testosterone propionate, during which time their feeding was recorded . Feeding variables were computed for each animal . (2d) The feeding pattern of mice growing rapidly at puberty. The feeding behaviour of eight male mice was recorded for 3 days during the rapid growth phase following puberty . Variables of feeding were computed, and compared with those of young and old castrates, and animals injected with testosterone using, in the main, Student's t-test . (3) The Effects of Female Company Upon the Feeding of Male Mice This was examined by a modification of the feeder cages . A light-gauge wire cage was placed within to occupy about one-third of the floor area, and provided with supplies of food and water. One animal was placed in the main cage and another in the inner one . The mesh size was such that the animals could enjoy through it a good deal of physical contact, but the beast in the inner cage could not reach the feeder recorder . In all cases a male lived in the main cage, and a female in the inner sanctum. (3a) Effects of female company upon food intake. The food intake of eight intact males was followed whilst they were caged with normal females showing oestrous cycles . The cycle of the female was monitored by taking daily vaginal smears, which were stained and scored for stage of cycle by a method described previously (Petersen 1976) . The food intake of each male on each day of the oestrous cycle of the female with whom he lived was collated until each pair had experienced five female cycles . Data were analysed by Friedman's test, and in addition a paired Student's t-test was used to compare the mean intake of the males over all days of the females cycle with their intake when living alone, which was measured for 7 days at the end of the experiment . (3b) Effects of female company on the temporal patterns of feeding . The temporal pattern of feeding of a further eight intact male mice was recorded whilst they were alone for 3 days, and whilst caged with females rendered periodically sexually receptive by the injection of 20 pg oestradiol benzoate every fourth day, a regime used by Manning (personal communication) for the induction of receptivity in female mice used

947

for behavioural testing of males . The feeding patterns of males with non-receptive and receptive females could then be compared using paired Students, t-tests . Results (1) Effects of Testosterone Upon Food Intake and Body Weight The effects on body weight of castration at different ages is shown in Fig . 1 . Pre-puberally castrated males weigh less than intact males at all ages after puberty (t-test P < 0 . 001, 14 df, comparison between intact and prepuberal castrates at 50 days of age . Similar statistics are significant at all ages after 40 days) . Castration in adulthood leads only to a slight slowing of weight gain (t-test P > 0 . 1, 22 df, comparison between intact and castrated males at 28 days post-operative) . Food intake figures reflect these changes, and are summarized in Table I . Testosterone injected into adult, pre-puberally castrated animals (experiment la) leads to a marked, and statistically significant rise in food intake (paired t-test P < 0 . 01, 7 df, comparing 30-

Body

weighs 20-

10

20

40

110 Age e.,

Fig . 1 . Body weight as a function of age in intact and castrated mice . Closed symbols intact animals, open symbols gonadectomized, triangular symbols male, circular female. For clarity, where groups are similar in weight both are represented by a single line . Table I. Food Intake of Intact and Castrated Males Food intake g/day Intact adult males (N = 12) Males castrated in adulthood (N = 12) Males castrated in adulthood and injected with testosterone (N = 6) Adult males castrated at weaning (N = 6) Adult males castrated at weaning (N = 8) and injected with testosterone

3 .5 ± 0 .1 3 .4 ± 0. 1 3 . 55 ± 0 . 1 3.1 ± 0.1 3 .8 ± 0.1

948

ANIMAL BEHAVIOUR, 26, 3

animals before and during injections, in each case using the mean intake over a 5-day period in each animal) . Treatment of males castrated in adulthood (experiment lb) has a much less marked effect upon intake, which is not significant (t-test P > 0 . 1, 10 df, comparison of injected and non-injected groups) . 2 . The Effects of Testosterone Upon Feeding Patterns Before discussing the specific results some mention needs to be made of the basic characteristics of the feeding pattern in these mice . This basic pattern is in all respects similar to that shown by females of the same strain, which has been discussed in detail elsewhere (Petersen 1976) . Briefly, animals eat in meals, each composed of a number of feeding bouts separated by relatively short intervals. The meals are themselves separated by relatively long intermeal intervals, defined as an interbout interval exceeding a certain criterial length established

Day

after consideration of the semi-logarithmic survivorship curve of all interbout intervals . The meals themselves are unevenly distributed over the day, being concentrated mainly in the hours of darkness. Even then, however, the meals are not spread evenly, being further concentrated into two or three periods of intense feeding (sessions usually occurring at the beginning and towards the end of the night) . Figure 2 is an idealized illustration of a typical pattern of meals in the mice used in this study . (2a) Effects of castration in adulthood upon feeding pattern. Castration significantly alters most variables of feeding in male mice . Meal size is significantly increased (paired t-test P < 0 .05, 7 df), comparing the eight males in experiment 2a before and after castration (see Fig . 3) . Table II also contains data recruited from other experiments in the series, allowing a .w' 5

Night sma IMI

I 1111

1111111111111111

Fig. 2. The basic pattern of meals in the mouse .

Fig . 3 . Male size and IMI (total and short class) before and after castration in the male mouse . Open histogram before, shaded after .

Table II. Feeding Pattern of Intact and Castrated Male Mice Meal size (s)

Intermeal interval (s)

No . bouts$ per meal

Mean bout duration (s)

Intact, adult male (N = 16)

51 . 5 ± 6 .7

3814 ± 161

14 . 1 ± 2 .0

3 .65 ± 0 . 9

Male castrated in adulthood (N = 16)

77 . 7 ± 9 .0

4138 ± 146

13 . 5 ± 1 . 8

5 . 75 ± 0 . 9

Adult male castrated at weaning (N = 6)

80 . 6 ± 7 .4

3944 ± 161

14 . 1 ± 2 .2

5 .71 ± 0 . 8

Adult male castrated at weaning, injected with testosterone (N = 6)

87 . 4 f 8 .8

3230 ± 154

21 . 8 + 2 .4

4 . 01 + 0 . 7

Young, intact, male just after puberty (N = 8)

90 . 2 ± 11 .0

2582 ± 126

21 . 9 ± 1 . 9

4 . 12 + 0 . 9

The meal size and IMI of intact adult males are both significantly less than those of castrates (P < 0 . 01 and 0 . 05 respectively) . Meal size does not differ significantly amongst the other groups . IMI is significantly (P < 0 .01) reduced by testosterone injections into adult males castrated as weanlings, and is significantly (P < 0 . 001) reduced by testosterone injections into adult males castrated as weanlings, and is significantly (P < 0 .001) less in young intact than old intact males . Young intact and adult weanling castrates show significantly elevated number of bouts per meal (P < 0 . 01 in both cases) . The mean bout duration is significantly higher in animals lacking either endogenous or injected testosterone (P < 0 .05) .



PETERSEN : TESTOSTERONE AND FEEDING IN MALE MICE

comparison between two groups of eight mice (in which case, unpaired t-test P < 0 . 01, 14 df ). Overall meal frequency. on the other hand is significantly reduced (paired t-test on experiment 2a data alone P < 0 . 05, 7 df. Using full data set unpaired t-test P < 0 .05,14 df ) . The reduction is nearly sufficient to compensate for the rise in meal size, so that overall the total time feeding does not change significantly (t-test P < 0 . 1, 14 df ) . The change in meal frequency is concentrated exclusively upon one class of meals, those falling within feeding sessions (where a session is defined by Petersen (1976) as a collection of meals separated by IMI's of less than 20 min) . The change in intrasession IMI is significant (paired t-test on 2a data P < 0 . 05, 7 df, unpaired t-test on all data P < 0 . 01, 14 df), whereas that in extra-session IMI's is not significant (paired t-test P > 0 . 1, 7 df, unpaired for total data P > 0- 1, 14 df ). The change in meal size affects all day and night meals . The mean number of bouts per meal is not significantly altered by castration (paired t-test on 2a data P > 0 . 1, 7 df, unpaired t-test on total data P > 0 . 1, 14 df) . Mean feeding bout length is, however, significantly altered (paired t-test P < 0 .05, for 2a data alone, P < 0 . 05 for total data, 14 df) . Castration alters the rate of increase of feeding bout length at the onset of the meal (a `positive feedback' process first described by Wiepkema (1971)). The third, fourth and fifth bouts are significantly longer in castrated animals than in their intact counterparts (Fig . 4) . (In each case

949

the mean duration of bouts in each position of meals containing at least eight bouts was calculated for each mouse, and t-tests used to compare figures before and after castration. Using only data from experiment 2a and paired t-tests for the first and second bouts P > 0 . 1, for the third P < 0 . 05, for the fourth P < 0 . 01, and for the fifth P < 0 . 05, with 7 df in each case . If all data are used the significance levels remain the same, except for bout three, where P < 0 . 01, with 14 df.) (2b) Effects of injected testosterone upon the feeding pattern of males castrated in adulthood . Testosterone produces a dose-responsive fall in meal size (Fig . 5), and a corresponding rise in meal frequency, which together combine to produce only a non-significant change in total time feeding. (Friedman analysis of variance across doses x = 19 . 7, P < 0 . 01 for meal size, x = 14 . 2, P < 0 . 01 for IMI, x = 2 . 3, P > 0 . 1 for total time feeding .) The changes are in all respects so far examined a simple reversal of the effects of castration so that, for example, changes in meal size affect all meals, whereas changes in meal frequency are concentrated on the intrasession class . (2c) Effects of injected testosterone upon the feeding patterns of adult, prepuberally castrated males. Here, the effects of testosterone are somewhat different . There is a slight nonsignificant rise in meal size (paired t-test P > 0- 1, 5 df), but meal frequency rises a great deal (paired t-test P < 0 . 01, 5 df) producing the overall rise in food intake described above . The

12-

70-

e ~e

I

60

8Mea 81ee

Bout Length

50-

4

20

30

40

Duse Ye

i

i

3

4

i

6

Bout

Fig. 4 . Initial facilitation of bout length before and after castration . (Symbols as in Fig. 1 .)

Fig. 5 . Dose response curve for effects of injected testosterone upon meal size in castrated male mice . In all cases the meal size change was accompanied by a change in IMI sufficient to prevent any significant alteration of food intake. The effects of dose were statistically significant (one way analysis of variance F = 5 . 2, P < 0 .01) .



95 0

ANIMAL BEHAVIOUR, 26, 3

number of bouts per meal rises significantly (paired t-test P < 0 .05, 5 df), so that despite the rise in meal size, the mean bout duration falls . (2d) Feeding of young mice . Males growing very rapidly just after puberty show a very high meal frequency, significantly larger than any class of males so far recorded (see Table II for statistics) . The meal size is also high, however, being as large as in castfate animals, but the number of bouts per medl is also very high so that overall, the mean bout duration is about that of an intact male (Table II) . (2e) Relationships between meal sizes and intermeal intervals. In rats (Le Magnen 1971) and mice (Petersen 1976) it is possible under some conditions to observe a correlation between meal size and following intermeal interval . In female mice this only occurs within pairs of meals contained in feeding sessions (defined as above), and even then is not particularly strong . The same is true for all classes of males recorded in these studies . A proportion of the males exhibit statistically significant correlations between meal size and following IMI, whereas no significant correlation between preceding meal size and IMI has ever been observed . It is possible to compute from such data a slope of regression between meal size and following IMI, although with only weak correlations it is not too clear how useful an exercise this might be . If it is 1200-

done, however, it appears that the slope of regression is unaffected by castration in adulthood, but is altered during the rapid growth phase following puberty, and after testosterone injections in adult, pre-pubertally castrated animals (Fig . 6) . (3) Effects of a Companion Upon Feeding (3a) Changes in food intake . The food intake of an intact male is significantly depressed when the female with whom he is caged comes on heat (Fig . 7) (over the whole female cycle Friedman's x for male intake = 16 . 7, P < 0 .01 ; comparing male intake when alone, and when with a receptive female, paired t-test, P < 0 . 01 ; 7 df; comparing male intake alone with intake with a non-receptive female, paired t-test P > 0 . 1, 7 df ) . If the mean intake of the male over all days of the female cycle is compared with his intake alone, however, there is no difference (paired t-test P > 0 . 1, 7 df) . (3b) Feeding pattern with a companion . Caging a male with a non-receptive female reduces his meal size and increases his meal frequency, but neither change is significant (paired t-test P > 0 . 1, 7 df in both cases) . When, however, the female is made receptive, there is a significant fall in meal size (paired t-test P < 0 . 01, 7 df), and a slight, but non-significant, rise in meal frequency, which fails to compensate the change in meal size, thus allowing the observed change in intake. The number of bouts per meal is unchanged, whereas the mean bout duration is significantly reduced (paired t-test P < 0 . 05, 7 df) . Figure 8 illustrates a comparison of meal 4-

800IMI

Intake of Male 3`m

400-

40

80 Meal Size

120

see

2-

Fig. 6. Best fit regression lines between meal size and following IMI within feeding sessions . Line A : intact male : line B : male castrated in adulthood ; line C : male castrated before puberty and injected with testosterone as an adult ; line D : young male during rapid growth phase following puberty .

i

1'

6

M

D

Stage of Female Cycle

Fig. 7 . The food intake of a male mouse caged with a female as a function of the oestrous cycle of the female .



PETERSEN : TESTOSTERONE AND FEEDING IN MALE MICE

sizes in male mice alone, after castration, and with receptive and non-receptive females . Discussion These results suggest that a case can be made for the existence of more than one influence of testosterone on feeding in the mouse . As in the rat, testosterone promotes weight gain after puberty, and this weight gain is associated with enhanced food intake . Despite, however, the fact that the male is larger, and eats more than either the female or a prepuberally castrated animal, his meals are smaller. This is not the case with the rat (Balagura & Devenport 1970) . Castration of the adult male produces only a slight fall in food intake, but this is associated with a marked rise in the size of meals . Injection of testosterone produces a dose responsive fall in meal size towards that of the intact male . These changes in meal size are accompanied by alterations of meal frequency which tend to oppose their effects upon food intake, thus producing the only relatively small changes which appear in this variable . The question remains, which of these changes, in meal size or frequency is primary? Le Magnen (1971) has suggested that, having observed a positive correlation between meal size and following IMI, meal size is the relatively independent variable of feeding, so that if such a correlation can also be demonstrated in these animals, then it is likely that the change in meal size is primary, and the changes in meal frequency is determined by it . In both female (Petersen 1976) and male mice (see above), it is possible to demonstrate such a positive correlation, but only between meal size and IMI pairs contained within feeding `sessions' at the beginning and end of the night . If, therefore, the changes in meal frequency are dependent upon those in meal size, only intrasession IMI's should be altered, as these are the

4 4-

-r4-

4o-

G.t, .

Intact

With NA,

111111, a'°4

Fig . 8 . Meal size of male mice after castration or caging with receptive or non-receptive female mice.

951

only ones which can be demonstrated to be dependent upon meal size. This is indeed the case . Furthermore, the slope of regression between meal size and IMI is not significantly altered by castration or testosterone therapy in adulthood, although, given the weakness of the correlations observed, it is not clear what significance can be attached to this observation . In adulthood, therefore, it would seem reasonable to suggest that testosterone influences primarily meal size, this having secondary repercussions upon meal frequency. Around puberty, testosterone promotes a rapid rate of weight gain, and food intake increases markedly . At this time meal size is raised, and meal frequency is higher than at any other time in the animal's normal life . The relationship between meal size and IMI is clearly altered in this case, so that a given meal size staves off the resumption of feeding for very much less time than it does in later life . Here then, the increased food intake due to testosterone is produced both by alteration of meal size (paradoxically in the opposite direction to that observed in adulthood) and meal frequency, the latter appearing by virtue of an alteration of the relationship between meal size and following IMI . Despite this higher meal size, the accompanying rise in the number of bouts per meal leads to a significantly lower mean bout duration than in intact castrates . Testosterone is therefore still reducing feeding bout duration and we need to examine whether the effect is the same as in the adult . The effect on adults of testosterone upon meal size does indeed come about mainly by alteration of the mean duration of feeding bouts within meals, and as is the case with the sexually receptive female (Petersen 1976), the main variable altered is the rate of initial facilitation of bout length in the opening stages of a meal . We must ask why meal size is altered in this way . It is difficult to see how a lowered meal size could be a specific adaptation to the higher food intake of the male, as when he is castrated his food intake falls only slightly, yet his meal size rises markedly. In addition, when his food intake is high, around puberty, meal size is raised . Clearly, therefore, meal size is altered for some reason not directly associated with the maintenance of energy balance. Castration produces a number of behavioural changes, including a much lowered tendency to indulge in sexual (McGill & Manning 1976) and aggressive behaviour . It is possible that the changes

952

ANIMAL BEHAVIOUR, 26, 3

in meal size are a direct or indirect consequence of alterations of those behaviours . Perhaps the propensity to perform sexual and aggressive behaviour affects the animal's behavioural systems in such a way as to alter meal size, just as occurs when the female is on heat (Petersen 1976). In the male, if such a `mood' exists, then it is continuously available for study, unlike female sexual receptivity, and further predictions may be examined . If a propensity to perform an activity in the absence of a suitable partner alters feeding, then it should be altered even more if a partner is provided . This was observed in the final experiment . Males caged with females show an additional alteration of feeding when the female is sexually receptive, which appears to be a simple extension of the difference between castrated and intact males . We seem to have demonstrated, therefore, an alteration of feeding which is relatively unrelated to the animal's metabolic state, and depends upon an interaction between internal, endocrine signals, determining the propensity to perform a behaviour, and external stimuli associated with that behaviour. This appeared quite separate from the changes in feeding which accompany the food intake changes during the rapid growth spurt promoted by testosterone . As it bears some general qualitative relationship to changes in feeding observed in female animals on heat,

we may be observing some general mechanism for altering feeding when some other activity has acquired a high priority . REFERENCES Balagura, S . & Devenport, L . D . 1970. Feeding patterns of normal and ventromedial hypothalamically lesioned male and female rats . J.C.P.P., 71, 357364 . Bell, D . D. & Zucker, I. 1971 . Sex differences in body weight and eating-organisation and activation by gonadal hormones in the rat. Physiol. & Behav., 7,27-34. Kakolewski, J . W ., Cox, V. C. & Valenstein, E. S . 1968 . Sex differences in body weight change following gonadectomy in male and female rats . Psychol. Rep., 22, 547-554. Le Magnen, J . 1971 . Control and regulation of food intake . Prog. in Physiol. Psych., 4, 203-261 . McGill, T . E . & Manning, A . 1976 . Genotype and retention of the ejaculatory reflex in castrated male mice. Anim. Behav., 24, 507-518 . Petersen, S . A. 1976. The temporal pattern of feeding over the oestrous cycle of the mouse . Anim. Behav., 24, 939-955 . Wade, G . N . 1972 . Gonadal hormones and the behavioural regulation of body weight . Physiol. & Behav., 8, 523-534 . Wiepkema, P . R. 1971 . Positive feedbacks at work during feeding. Behaviour, 35, 266-273 . Wright, P. & Turner, C . 1973 . Sex differences in body weight following gonadectomy and goldthioglucose injections in mice. Physiol. & Behav ., 11, 155-159 . (Received 7 July 1977 ; revised 18 August 1977 ; MS. number : 1652)