Anim. Behav ., 1975, 23, 773-778 Er ECTS OF NEONATAL CASTRATION AND TESTOSTERONE INJECTION ON ADULT OPEN-FIELD BEHAVIOUR IN RATS WITH ATYPICAL SEX DIFFERENCE IN DEFECATION BY JEFFREY A . GRAY, ROBERT F . DREWETT* & BARBARA LALLJEE Department of Experimental Psychology, University of Oxford Abstract. In the open-field test, male rats usually defecate more and ambulate less than females . A strain was studied in which the males defecate less than females, while still ambulating less . Infants of this strain were castrated (males) or injected with testosterone propionate (females) and tested in the open field as adults . There were significant effects of these neonatal treatments on ambulation, but not on defecation, in contrast to previous reports for the latter measure in rats showing a sex difference in defecation taking the usual direction. It is suggested that absence of the usual direction of a sex difference normally under the developmental control of androgen may indicate a genotype which has escaped from such control . Finally, a number of `sibling' effects' were observed, i .e. effects on the adult behaviour of the members of one sex in a litter produced by treatments administered in infancy to members of the other sex . normal sex difference in open-field defecation is absent (females obtaining higher scores than males) . They also present evidence that this reversal of the sex difference in defecation has been accompanied by a more general reversal of sex differences in emotional behaviour in this strain. In the present investigation, we set out to answer the following question : is open-field defecation in the Oxford hooded strain influenced by neonatal testosterone as it is in those strains showing the more usual sex difference? We have examined the effects in this strain of injections of testosterone to the infant female and castration of the infant male on adult defecation in the open field . In addition, we investigated ambulation in the open field, a measure on which the Oxford hooded rats show a sex difference of the usual kind (females ambulating more).
Male rats normally defecate in the open field more than females, and also obtain higher scores on a number of other presumed tests of fearfulness (Gray 1971) . This sex difference in emotional behaviour has regularly appeared in outbred strains of rats . However, the sex difference in open-field defecation seems to disappear or even be reversed in strains of rats which have been selectively bred (and at the same time inbred) on criteria connected with emotionality (Sines 1961 ; Gray, Levine & Broadhurst 1965) or in rats which have simply been inbred (Gray & Lalljee 1974). The same pattern appears in the mouse : outbred animals show a sex difference of the same direction as that reported for the outbred rat, males defecating more than females, but inbred strains show this sex difference less or not at all (Bruell 1969) . When the sex difference in open-field defecation takes the usual direction in the rat, it responds to the same neonatal manipulations which affect other sex differences in endocrine processes and sexual behaviour (Gorski 1971) . Castration of the infant male rat decreases adult defecation scores (Pfaff & Zigmond 1971) and injections of testosterone propionate to the infant female rat give rise to an elevation in adult defecation scores, i.e. a change in the usual male direction (Gray et al . 1965 ; Gray, Lean & Keynes 1969 ; Drach et al ., cited by Valenstein 1968 ; Pfaff & Zigmond 1971) . Gray & Lalljee (1974) have recently described a strain of inbred hooded rats in which the *Present address : Department of Psychology, University of Durham.
General Method Subjects were drawn from the Oxford colony of hooded rats (Gray & Lalljee 1974) . After a period of about twelve generations of unsystematic inbreeding, this strain had been changed to a random mating r6gime, though no admixture of genes from outside the colony occurred . The experiments were carried out on animals of the third, fourth and fifth generations after random mating began. The maintenance of the litters used in the experiments followed essentially the pattern described by Broadhurst (1960), except that weaning was carried out at 28 instead of 21 days . 773
774
ANIMAL BEHAVIOUR, 23, 4
Infant Treatments Two main infant treatments were used : castration of male pups and injection of testosterone propionate (TP). TP was injected in a dose of 0 .5 mg in a vehicle of 0 .05 ml of a mixture of 20 per cent ethyl oleate and 80 per cent arachis oil subcutaneously via a 27-g needle inserted beneath the skin of the back . Placebo injections were 0 .05 ml arachis oil . Injections were made on day I of postnatal life (within the first 24 hr) or on day 5 (96 to 120 hr after birth). Removal of the testes was accomplished via a ventral mid-line incision, using cooling on ice as the anaesthetic . Sham operation, in both males and females, consisted of the insertion of forceps through the incision . Wounds were closed by a drop of coloidon . All surgery was conducted on day 1 of life . For all treatments, whole litters were removed and the pups were sexed and weighed . The litter was returned to the home cage at the end of the appropriate treatment . Untreated control litters were removed from the home cage, sexed and weighed, and left in a wooden box covered in cotton wool for a time equivalent to the length
of the treatment for which they served as controls . This procedure ('handling') was carried out on day 1 or on day 5 of life . To increase the likelihood that the mother would accept the returned pups, no matter what treatment they had had, she was briefly exposed to ether vapour before they were replaced . Open-Field Testing Testing was conducted over four successive days when the rats were approximately 100 days old. The OF test was closely modelled on the one described by Broadhurst (1960 ; and see Gray et al. 1969) . Illumination level was 1614 lux and noise level 78 dB . Ambulation (A) and defecation (D) scores were recorded for each of four consecutive daily 2-min trials . The A score is reported as number of compartments (marked on the floor of the arena) entered and may be converted to metres by dividing lay 4 .63 . The D score is reported as number of faecal bolusses dropped . Owing to the large number of zero scores on single days, in experiments 1 and 2 defecation was analysed only as the total over the 4 days of testing.
Table I. Experimental Designs. (TP : Testosterone Propionate. N : Number of Animals of Each Sex) Treatment Experiment 1
2
3
Group
Day of Treatment
Male
Female
N
Normal 1 (NI)
1
Handling
Handling
15
Normal 5 (N5)
5
Handling
Handling
8
Placebo 1(Pl)
1
Placebo
Placebo
21
Placebo 5 (P5)
5
Placebo
Placebo
15
Sham-operation (SI)
1
Sham-operation
Sham-operation
8
Placebo 1(P1)
1*
Placebo
Placebo
7
Placebo 5 (P5)
5•
Placebo
Placebo
7
TP I (Ti) TP 5 (T5)
1
Placebo
TP
7
5 1
Placebo
TP
7
Sham-operation
Placebo
8
Castrate-placebo (C-P) Sham-TP (S-T)
I
Castration
Placebo
8
Sham-operation
TP
8
Castrate-TP (C-T)
1
Castration
TP
8
Sham-placebo (S-P)
B oth the P1 and the P5 group of experiment 2 were also included in the corresponding groups of experiment 1 .
GRAY ET AL . : CASTRATION, . TESTOSTERONE AND OPEN FIELD TEST Experimental Designs The three experiments reported may be defined in terms of the infant treatments applied, as shown in Table I . Experiment 1 reports the sex differences found and their stability under control treatments . The P1 group of experiment 1 (Table I) was initially a separate experiment on the effects of varying the temperature of the oil placebo, there being three subgroups (each of seven males and seven females) receiving 10, 24 and 35 °C injections respectively . No effects of this variation in temperature were found. The PI group of experiment 2 consists of the sub-group injected at 24°C, i .e. approximately room temperature. Experiment 2 investigates the effects of injections of TP to infant females, and experiment 3 the effects of this treatment and that of castration of infant males . Rats were randomly selected for adult testing from the surviving members of the infanttreated litters so as to provide seven to eight subjects of each sex, drawn from two to three separate litters, for each of the experimental conditions in experiments 2 and 3 (Table I) . Both of these experiments are designed to take account of sibling effects of the kind observed by Gray et al . (1969), i .e. effects on animals of one sex within a litter produced by treatments applied to their opposite-sexed siblings . The results of all three experiments were subjected to analyses of variance and subsequent t-tests based on the appropriate error terms from these analyses . Results Experiment 1 The observed sex differences confirmed those reported by Gray & Lalljee (1974) . Males defecated significantly less than females (F = 24 .4, df = 1/116, P < 0 .001) ; the male mean was 1 .0+ (SE) 0 .11 bolus per day, the female mean 1 .7 ± 0 .15 . Males also ambulated less than females (F = 5 .4, df = 1/116, P < 0 .05) ; the male mean was 20 .0 ± 0 .9 compartments per day, the female mean 24 .1 ± 0-9 . On neither the D nor the A score was there an interaction between sex and the five different control treatments listed in Table I (F = 1 .9 and 0 .6 respectively, df = 8/116) . (There were effects of the control treatments themselves, of the same kind in both sexes, but these will not be reported .) Experiment 2 There was no effect of the infant TP treatment on D scores in the females (Table II) . Sex
775
Table 11 . Effect of Infant TP on D Score In Experiment 2 . Mean Boluses per Day ± SE . Groups as in Table L Group
Males .32 ± 0 .32 1
P1 T1
2 .04 ± 0 .33 2 .32 f 0 .20
T5
1 .43 ± 0 .42
P5
40
0° a---o Placebo --- Placebo
Females .61 f 0.21 2 2 .29 f 0.46 2 .79 ± 0 .32 2 .79 f 0 .30
9 o-o Placebo H Testosterone
Q'
10 01
1
2 3 Days in Open Field
Fig. 1 . Effects of infant TP or placebo injection on adult ambulation scores as a function of sex and day of open field-testing in experiment 2. remained highly significant (F = 12 .4, df = 1/48, P < 0 .01), females defecating more than males as in experiment 1 (Table II) . Analysis of the A scores showed a significant interaction between treatment and day of testing (F = 2 .66, df = 3/144, P = 0 .05), uncomplicated by further interaction with sex or day of treatment . This interaction is shown in Fig . 1 . from which it can be seen that infant TP reduced adult ambulation on days 1, 2 and 4 of OF testing (P < 0 .01 in each case by t-test) . The effects of TP were very comparable whether the injection took place on day I or day 5 of life . Furthermore they were almost as evident among the oil-injected male siblings in the Tl
77 6
ANIMAL BEHAVIOUR, 23, 4
and T5 groups as among the hormone-injected females (Fig . 1). Experiment 3 The data from this experiment were submitted to analyses of variance in which the betweensubjects factors were sex, treatment (experimental or control, i .e . castration and TP versus sham-operation and placebo), and siblings' treatment (experimental or control, i.e. siblings' castration and TP versus siblings' shamoperation and placebo) . In addition, the withinsubjects factor of day of OF testing was included . The analysis of the D scores revealed no significant effect of either castration in males or TP injection in females (Table III) . Sex remained highly significant (F = 9 .9 ; df = 1/56, P < 0 .01), females defecating more than males as expected (Table III) . However, males and females received different treatments, so this finding does not necessarily represent a pure sex difference . Analysis of the A scores revealed a significant interaction between treatment and sex (F = 9 .25, df = 1/56, P < 0 .01), as well as a highly complex interaction between these two factors, siblings' treatment and day of OF testing (F = 4 .92, df = 3/168, P < 0 .01) . These interactions are displayed in Figs . 2 and 3 . Application of the appropriate t-tests to the two-way interaction showed that the effect of castration of the male, considered irrespective of siblings' treatment or day of OF testing, was not significant, whereas the effect of TP injection to the female was (P < 0 .01) . When applied to the four-way interaction, the t-tests showed that the effects of both castration of the male and of TP injection to the female on the directly treated animals depended in part on the treatment accorded to their respective siblings. As shown in Fig. 2, the overall effect of castration was to increase ambulation. But (Fig. 3) this Table III. Effect of Infant Castration or TP on DDScore in Experiment 3 . Mean Boluses per day f SE. Groups as in Table I Group C-P
Males Females .59 ±0 .37 2.12 ±0 .18 1 1 .44 f 0 .27 1 .75 f 0 .27
S-T
1 .03 ± 0 .23 2.34 f 0 .28
S-P
C-T
1 .72 ± 0 .37
2.28 f 0 .35
effect was significant (P < 0 .05) on days 3 and 4 of OF testing in the group of males whose sisters had been placebo-injected, and on days 2 and 4 (P < 0 .01) in the group whose sisters had had TP. The overall effect of TP, in contrast, was to decrease ambulation (Fig. 2), a decrease which was significant on days 2 (P < 0 .001) and 3 (P < 0 .01) of OF testing in the group of females whose brothers had received sham operations but on days 1 and 4 (P < 0 .01) in the group whose brothers had been castrated (Fig . 3). Discussion The pattern of sex differences reported for the Oxford hooded rats by Gray & Lalljee (1974) was confirmed in experiment 1 . Females defecated more in the open field than males ; this difference is the reverse of that usually found in the rat (Gray 1971). In the same test situation, females ambulated more than males ; the direction of this sex difference is the usual one found in the rat (Gray 1971). The positive effects of neonatal castration and TP injection observed in experiments 2 and 3 were in line with previous reports and form a consistent pattern. Thus injection of TP to females, whether on day 1 or day 5 of life, reduced the adult A score (i .e . a change in the male direction), confirming the findings of Gray et al . (1965, 1969), Pfaff & Zigmond (1971), Swanson (1967), Drach et al. (cited by Valenstein 1968), Gorski & Quadagno (cited by Gorski 1971) and Blizard & Denef (1973) . Conversely, castration
20 N 0 N C0
dr
9
10-
ho
T 0 0A I Sham Castrate
Placebo Testo -sterone Fig 2. Effect of infant castration in males and infant TP in females on open-field ambulation (mean ± sE) in experiment 3 .
GRAY ET AL. : CASTRATION, TESTOSTERONE AND OPEN FIELD TEST
d U,
20
0-0 0---1 o--o 6-9
Males 0~' Y Sham Placebo Castrate Placebo Sham Testosterone Castrate Testosterone
777
Femates 4 o-o Sham Placebo o--o Castrate Placebo o--o Sham Testosterone r-o Castrate Testosterone 0
•
o
~~
0 ,
A
o_
\`o 0
1
2
3
1 4 Days in Open Field
2
3
4
Fig . 3 . Interactive effects on ambulation of sex, treatment, siblings' treatment and day of open-field testing in experiment 3 . of the infant male increased adult ambulation, as previously found by Pfaff & Zigmond (1971), this being a change in the female direction . It is to be noted, however, that these positive results are confined to the ambulation measure . What is unusual about the present findings is the lack of effect of any of the infant treatments on open-field defecation. In other strains of rats infant injections of TP to females have been reported to increase D scores in the open field (Gray et al . 1965, 1969 ; Drach et al ., cited by Valenstein 1968 ; Pfaff & Zigmond 1971) and castration of the infant male to decrease these scores (Pfaff & Zigmond 1971) . It is not, therefore, likely that our results are due to any general insensitivity of OF defecation to neonatal testosterone . Nor can they be attributed to some general ineffectiveness of our neonatal procedures, given the positive findings on ambulation in the same animals . It seems reasonable to conclude, therefore, that in the Oxford hooded strain OF ambulation is affected by variation in neonatal testosterone levels, but OF defecation is not . As noted above, the Oxford hooded rats display a sex difference in OF defecation which is the reverse of that usually found in the rat . It is possible that the resistance of OF defecation
to manipulation of neonatal gonadal hormones is connected with this reversal of the sex difference. Support for this hypothesis can be drawn from a detailed comparison of the results of the present study and those reported by Gray et al . (1965, 1969) . Gray et al . (1969) used an unselected Wistar population with a large sex difference in the usual direction (males defecating more) and found a substantial effect of infant TP treatment (the female score being elevated) . In the Gray et al . (1965) experiment, a mixed group of Wistar-derived Maudsley Reactive (MR : Broadhurst 1960) rats and the F1 derived from a cross between these and Maudsley Non-reactive rats was used. The M R rats displayed no sex difference of any kind and the F 1 a sex difference of the usual kind . The effect of infant testosterone injections to the females of this mixed group was of the same kind as that observed in the unselected Wistaas used by Gray et al . (1969), but it was much smaller and of only marginal significance . In the present experiment, the sex difference in defecation was the reverse of usual, and the neonatal treatments had no effect at all . Thus, on the basis of all three studies, it is possible provisionally to conclude that, the greater the sex difference in defecation in the usual direction for the
778
ANIMAL BEHAVIOUR, 23, 4
rat (males defecating more), the greater is the dependence of the sex difference on the normal processes of sexual differentiation . A minor aspect of the results which deserves comment is the presence of what Gray et al . (1969) have termed `sibling effects', i .e. an effect on the adult behaviour of the siblings of one sex in a litter produced by a treatment given to the siblings in the same litter belonging to the opposite sex . Thus, in experiment 2, males whose sisters were injected with TP on day 1 had significantly (P < 0 .05 by t-test based on analysis of variance) higher D scores than males whose sisters received placebo on that day (see Table II), and also significantly lower A scores (P < 0 .01 ; Fig . 1) ; although the males received oil in both cases. Similarly, in experiment 3, as shown in Fig . 3, males whose sisters were injected with TP returned significantly (P
Male rats normally defecate in the open field more than females, and also obtain higher scores on a number of other presumed tests of fearfulness (Gray 1971) . This sex difference in emotional behaviour has regularly appeared in outbred strains of rats . However, the sex difference in open-field defecation seems to disappear or even be reversed in strains of rats which have been selectively bred (and at the same time inbred) on criteria connected with emotionality (Sines 1961 ; Gray, Levine & Broadhurst 1965) or in rats which have simply been inbred (Gray & Lalljee 1974). The same pattern appears in the mouse : outbred animals show a sex difference of the same direction as that reported for the outbred rat, males defecating more than females, but inbred strains show this sex difference less or not at all (Bruell 1969) . When the sex difference in open-field defecation takes the usual direction in the rat, it responds to the same neonatal manipulations which affect other sex differences in endocrine processes and sexual behaviour (Gorski 1971) . Castration of the infant male rat decreases adult defecation scores (Pfaff & Zigmond 1971) and injections of testosterone propionate to the infant female rat give rise to an elevation in adult defecation scores, i.e. a change in the usual male direction (Gray et al . 1965 ; Gray, Lean & Keynes 1969 ; Drach et al ., cited by Valenstein 1968 ; Pfaff & Zigmond 1971) . Gray & Lalljee (1974) have recently described a strain of inbred hooded rats in which the *Present address : Department of Psychology, University of Durham.
General Method Subjects were drawn from the Oxford colony of hooded rats (Gray & Lalljee 1974) . After a period of about twelve generations of unsystematic inbreeding, this strain had been changed to a random mating r6gime, though no admixture of genes from outside the colony occurred . The experiments were carried out on animals of the third, fourth and fifth generations after random mating began. The maintenance of the litters used in the experiments followed essentially the pattern described by Broadhurst (1960), except that weaning was carried out at 28 instead of 21 days . 773
774
ANIMAL BEHAVIOUR, 23, 4
Infant Treatments Two main infant treatments were used : castration of male pups and injection of testosterone propionate (TP). TP was injected in a dose of 0 .5 mg in a vehicle of 0 .05 ml of a mixture of 20 per cent ethyl oleate and 80 per cent arachis oil subcutaneously via a 27-g needle inserted beneath the skin of the back . Placebo injections were 0 .05 ml arachis oil . Injections were made on day I of postnatal life (within the first 24 hr) or on day 5 (96 to 120 hr after birth). Removal of the testes was accomplished via a ventral mid-line incision, using cooling on ice as the anaesthetic . Sham operation, in both males and females, consisted of the insertion of forceps through the incision . Wounds were closed by a drop of coloidon . All surgery was conducted on day 1 of life . For all treatments, whole litters were removed and the pups were sexed and weighed . The litter was returned to the home cage at the end of the appropriate treatment . Untreated control litters were removed from the home cage, sexed and weighed, and left in a wooden box covered in cotton wool for a time equivalent to the length
of the treatment for which they served as controls . This procedure ('handling') was carried out on day 1 or on day 5 of life . To increase the likelihood that the mother would accept the returned pups, no matter what treatment they had had, she was briefly exposed to ether vapour before they were replaced . Open-Field Testing Testing was conducted over four successive days when the rats were approximately 100 days old. The OF test was closely modelled on the one described by Broadhurst (1960 ; and see Gray et al. 1969) . Illumination level was 1614 lux and noise level 78 dB . Ambulation (A) and defecation (D) scores were recorded for each of four consecutive daily 2-min trials . The A score is reported as number of compartments (marked on the floor of the arena) entered and may be converted to metres by dividing lay 4 .63 . The D score is reported as number of faecal bolusses dropped . Owing to the large number of zero scores on single days, in experiments 1 and 2 defecation was analysed only as the total over the 4 days of testing.
Table I. Experimental Designs. (TP : Testosterone Propionate. N : Number of Animals of Each Sex) Treatment Experiment 1
2
3
Group
Day of Treatment
Male
Female
N
Normal 1 (NI)
1
Handling
Handling
15
Normal 5 (N5)
5
Handling
Handling
8
Placebo 1(Pl)
1
Placebo
Placebo
21
Placebo 5 (P5)
5
Placebo
Placebo
15
Sham-operation (SI)
1
Sham-operation
Sham-operation
8
Placebo 1(P1)
1*
Placebo
Placebo
7
Placebo 5 (P5)
5•
Placebo
Placebo
7
TP I (Ti) TP 5 (T5)
1
Placebo
TP
7
5 1
Placebo
TP
7
Sham-operation
Placebo
8
Castrate-placebo (C-P) Sham-TP (S-T)
I
Castration
Placebo
8
Sham-operation
TP
8
Castrate-TP (C-T)
1
Castration
TP
8
Sham-placebo (S-P)
B oth the P1 and the P5 group of experiment 2 were also included in the corresponding groups of experiment 1 .
GRAY ET AL . : CASTRATION, . TESTOSTERONE AND OPEN FIELD TEST Experimental Designs The three experiments reported may be defined in terms of the infant treatments applied, as shown in Table I . Experiment 1 reports the sex differences found and their stability under control treatments . The P1 group of experiment 1 (Table I) was initially a separate experiment on the effects of varying the temperature of the oil placebo, there being three subgroups (each of seven males and seven females) receiving 10, 24 and 35 °C injections respectively . No effects of this variation in temperature were found. The PI group of experiment 2 consists of the sub-group injected at 24°C, i .e. approximately room temperature. Experiment 2 investigates the effects of injections of TP to infant females, and experiment 3 the effects of this treatment and that of castration of infant males . Rats were randomly selected for adult testing from the surviving members of the infanttreated litters so as to provide seven to eight subjects of each sex, drawn from two to three separate litters, for each of the experimental conditions in experiments 2 and 3 (Table I) . Both of these experiments are designed to take account of sibling effects of the kind observed by Gray et al . (1969), i .e. effects on animals of one sex within a litter produced by treatments applied to their opposite-sexed siblings . The results of all three experiments were subjected to analyses of variance and subsequent t-tests based on the appropriate error terms from these analyses . Results Experiment 1 The observed sex differences confirmed those reported by Gray & Lalljee (1974) . Males defecated significantly less than females (F = 24 .4, df = 1/116, P < 0 .001) ; the male mean was 1 .0+ (SE) 0 .11 bolus per day, the female mean 1 .7 ± 0 .15 . Males also ambulated less than females (F = 5 .4, df = 1/116, P < 0 .05) ; the male mean was 20 .0 ± 0 .9 compartments per day, the female mean 24 .1 ± 0-9 . On neither the D nor the A score was there an interaction between sex and the five different control treatments listed in Table I (F = 1 .9 and 0 .6 respectively, df = 8/116) . (There were effects of the control treatments themselves, of the same kind in both sexes, but these will not be reported .) Experiment 2 There was no effect of the infant TP treatment on D scores in the females (Table II) . Sex
775
Table 11 . Effect of Infant TP on D Score In Experiment 2 . Mean Boluses per Day ± SE . Groups as in Table L Group
Males .32 ± 0 .32 1
P1 T1
2 .04 ± 0 .33 2 .32 f 0 .20
T5
1 .43 ± 0 .42
P5
40
0° a---o Placebo --- Placebo
Females .61 f 0.21 2 2 .29 f 0.46 2 .79 ± 0 .32 2 .79 f 0 .30
9 o-o Placebo H Testosterone
Q'
10 01
1
2 3 Days in Open Field
Fig. 1 . Effects of infant TP or placebo injection on adult ambulation scores as a function of sex and day of open field-testing in experiment 2. remained highly significant (F = 12 .4, df = 1/48, P < 0 .01), females defecating more than males as in experiment 1 (Table II) . Analysis of the A scores showed a significant interaction between treatment and day of testing (F = 2 .66, df = 3/144, P = 0 .05), uncomplicated by further interaction with sex or day of treatment . This interaction is shown in Fig . 1 . from which it can be seen that infant TP reduced adult ambulation on days 1, 2 and 4 of OF testing (P < 0 .01 in each case by t-test) . The effects of TP were very comparable whether the injection took place on day I or day 5 of life . Furthermore they were almost as evident among the oil-injected male siblings in the Tl
77 6
ANIMAL BEHAVIOUR, 23, 4
and T5 groups as among the hormone-injected females (Fig . 1). Experiment 3 The data from this experiment were submitted to analyses of variance in which the betweensubjects factors were sex, treatment (experimental or control, i .e . castration and TP versus sham-operation and placebo), and siblings' treatment (experimental or control, i.e. siblings' castration and TP versus siblings' shamoperation and placebo) . In addition, the withinsubjects factor of day of OF testing was included . The analysis of the D scores revealed no significant effect of either castration in males or TP injection in females (Table III) . Sex remained highly significant (F = 9 .9 ; df = 1/56, P < 0 .01), females defecating more than males as expected (Table III) . However, males and females received different treatments, so this finding does not necessarily represent a pure sex difference . Analysis of the A scores revealed a significant interaction between treatment and sex (F = 9 .25, df = 1/56, P < 0 .01), as well as a highly complex interaction between these two factors, siblings' treatment and day of OF testing (F = 4 .92, df = 3/168, P < 0 .01) . These interactions are displayed in Figs . 2 and 3 . Application of the appropriate t-tests to the two-way interaction showed that the effect of castration of the male, considered irrespective of siblings' treatment or day of OF testing, was not significant, whereas the effect of TP injection to the female was (P < 0 .01) . When applied to the four-way interaction, the t-tests showed that the effects of both castration of the male and of TP injection to the female on the directly treated animals depended in part on the treatment accorded to their respective siblings. As shown in Fig. 2, the overall effect of castration was to increase ambulation. But (Fig. 3) this Table III. Effect of Infant Castration or TP on DDScore in Experiment 3 . Mean Boluses per day f SE. Groups as in Table I Group C-P
Males Females .59 ±0 .37 2.12 ±0 .18 1 1 .44 f 0 .27 1 .75 f 0 .27
S-T
1 .03 ± 0 .23 2.34 f 0 .28
S-P
C-T
1 .72 ± 0 .37
2.28 f 0 .35
effect was significant (P < 0 .05) on days 3 and 4 of OF testing in the group of males whose sisters had been placebo-injected, and on days 2 and 4 (P < 0 .01) in the group whose sisters had had TP. The overall effect of TP, in contrast, was to decrease ambulation (Fig. 2), a decrease which was significant on days 2 (P < 0 .001) and 3 (P < 0 .01) of OF testing in the group of females whose brothers had received sham operations but on days 1 and 4 (P < 0 .01) in the group whose brothers had been castrated (Fig . 3). Discussion The pattern of sex differences reported for the Oxford hooded rats by Gray & Lalljee (1974) was confirmed in experiment 1 . Females defecated more in the open field than males ; this difference is the reverse of that usually found in the rat (Gray 1971). In the same test situation, females ambulated more than males ; the direction of this sex difference is the usual one found in the rat (Gray 1971). The positive effects of neonatal castration and TP injection observed in experiments 2 and 3 were in line with previous reports and form a consistent pattern. Thus injection of TP to females, whether on day 1 or day 5 of life, reduced the adult A score (i .e . a change in the male direction), confirming the findings of Gray et al . (1965, 1969), Pfaff & Zigmond (1971), Swanson (1967), Drach et al. (cited by Valenstein 1968), Gorski & Quadagno (cited by Gorski 1971) and Blizard & Denef (1973) . Conversely, castration
20 N 0 N C0
dr
9
10-
ho
T 0 0A I Sham Castrate
Placebo Testo -sterone Fig 2. Effect of infant castration in males and infant TP in females on open-field ambulation (mean ± sE) in experiment 3 .
GRAY ET AL. : CASTRATION, TESTOSTERONE AND OPEN FIELD TEST
d U,
20
0-0 0---1 o--o 6-9
Males 0~' Y Sham Placebo Castrate Placebo Sham Testosterone Castrate Testosterone
777
Femates 4 o-o Sham Placebo o--o Castrate Placebo o--o Sham Testosterone r-o Castrate Testosterone 0
•
o
~~
0 ,
A
o_
\`o 0
1
2
3
1 4 Days in Open Field
2
3
4
Fig . 3 . Interactive effects on ambulation of sex, treatment, siblings' treatment and day of open-field testing in experiment 3 . of the infant male increased adult ambulation, as previously found by Pfaff & Zigmond (1971), this being a change in the female direction . It is to be noted, however, that these positive results are confined to the ambulation measure . What is unusual about the present findings is the lack of effect of any of the infant treatments on open-field defecation. In other strains of rats infant injections of TP to females have been reported to increase D scores in the open field (Gray et al . 1965, 1969 ; Drach et al ., cited by Valenstein 1968 ; Pfaff & Zigmond 1971) and castration of the infant male to decrease these scores (Pfaff & Zigmond 1971) . It is not, therefore, likely that our results are due to any general insensitivity of OF defecation to neonatal testosterone . Nor can they be attributed to some general ineffectiveness of our neonatal procedures, given the positive findings on ambulation in the same animals . It seems reasonable to conclude, therefore, that in the Oxford hooded strain OF ambulation is affected by variation in neonatal testosterone levels, but OF defecation is not . As noted above, the Oxford hooded rats display a sex difference in OF defecation which is the reverse of that usually found in the rat . It is possible that the resistance of OF defecation
to manipulation of neonatal gonadal hormones is connected with this reversal of the sex difference. Support for this hypothesis can be drawn from a detailed comparison of the results of the present study and those reported by Gray et al . (1965, 1969) . Gray et al . (1969) used an unselected Wistar population with a large sex difference in the usual direction (males defecating more) and found a substantial effect of infant TP treatment (the female score being elevated) . In the Gray et al . (1965) experiment, a mixed group of Wistar-derived Maudsley Reactive (MR : Broadhurst 1960) rats and the F1 derived from a cross between these and Maudsley Non-reactive rats was used. The M R rats displayed no sex difference of any kind and the F 1 a sex difference of the usual kind . The effect of infant testosterone injections to the females of this mixed group was of the same kind as that observed in the unselected Wistaas used by Gray et al . (1969), but it was much smaller and of only marginal significance . In the present experiment, the sex difference in defecation was the reverse of usual, and the neonatal treatments had no effect at all . Thus, on the basis of all three studies, it is possible provisionally to conclude that, the greater the sex difference in defecation in the usual direction for the
778
ANIMAL BEHAVIOUR, 23, 4
rat (males defecating more), the greater is the dependence of the sex difference on the normal processes of sexual differentiation . A minor aspect of the results which deserves comment is the presence of what Gray et al . (1969) have termed `sibling effects', i .e. an effect on the adult behaviour of the siblings of one sex in a litter produced by a treatment given to the siblings in the same litter belonging to the opposite sex . Thus, in experiment 2, males whose sisters were injected with TP on day 1 had significantly (P < 0 .05 by t-test based on analysis of variance) higher D scores than males whose sisters received placebo on that day (see Table II), and also significantly lower A scores (P < 0 .01 ; Fig . 1) ; although the males received oil in both cases. Similarly, in experiment 3, as shown in Fig . 3, males whose sisters were injected with TP returned significantly (P
