Serum testosterone, male dominance, and aggression in captive groups of vervet monkeys (Cercopithecus aethiops sabaeus)

HORMONES

AND

BEHAVIOR

19, 154-163 (1985)

Serum Testosterone, Male Dominance, and Aggression in Captive Groups of Vervet Monkeys (Cercopithecus Aethiops Sabaeus) HORST

D. STEKLIS,*.$ GARY L. BRAMMER,~*$*§ MICHAEL J. RALEIGH,+$‘~ AND MICHAEL T. Mc&mud”~

*Department of Anthropology, Douglass College, Rutgers University, New Brunswick, New Jersey 08903; tDepartment of Psychiatry and Biobehavioral Sciences, School of Medicine, University of California. Los Angeles, California 90024: #Nonhuman Primate Laboratory, Sepulveda Veterans Administration Medical Center, Sepulveda. California 91343; and 5Neurobiochemistry Laboratory, Brentwood Veterans Administration Medical Center, Los Angeles, California 90073

The relationship of serum testosterone concentration to male dominance rank and frequency of aggression was investigated in stable vervet monkey social groups, each containing two or three adult males, several adult females, and their offspring. Dominance relationships were determined by noting an animal’s success in intermale aggressive encounters. A striking finding was the marked within-subject variation in testosterone concentration: 5- to IO-fold fluctuations were often observed on successive days. When all I5 groups were considered together, testosterone concentration was unrelated to dominance rank. Although mean testosterone concentration for all dominant males was higher than the mean for all subordinate males, this difference was not significant. In a subset of 4 groups, the rate of aggression initiated was significantly correlated with same-day testosterone in dominant but not in subordinate males. ,% 1985 Academic Pmr.

Inc.

Examinations of the relationship between concentrations of peripheral testosterone (T) and dominance rank and aggressionin adult male nonhuman primates have yielded conflicting results (see Dixson, 1980; Bouissou, 1983for reviews). For example, plasma T concentrations were positively associated with rank in captive all-male groups of rhesus macaques (Rose, Holaday, and Bernstein, 1971)and squirrel monkeys (Coe, Smith, Mendoza, and Levine 1983), as well as in heterosexual groups of squirrel monkeys (Coe et al., 1983) and talapoin monkeys (Eberhart, Keverne, and Meller, 1980). In contrast, there was no correlation between plasma T levels and dominance rank in captive all-male talapoin groups (Eberhart et al., 1980), rhesus macaques (Gordon, Rose, and Bernstein, 1976), or freeranging baboons (Sapolsky, 1982). I54 ool8-506x/85 $1.50 Copyright All rights

G 1985 by Academic Press. Inc. of reproduction in any form reserved

SERUM

TESTOSTERONE

AND AGGRESSION

155

Some studies have reported that dominance rank and the frequency of aggression initiated are positively related (Rose et al., 1971; Coe et al., 1983). This observation suggests that the primary relationship is between T concentration and aggressivity with T and dominance being secondarily related. This suggestion is supported by the observations that in captive male stump-tailed macaques and free-ranging baboons aggressive behavior was positively related to T concentrations (Kling and Dunne, 1976; Sapolsky, 1982). Similarly in rhesus and pigtailed macaques, winning intermale aggressive encounters increases and losing decreases T concentration (Rose, Bernstein, and Gordon, 1975;Bernstein, Rose, Gordon, and Grady, 1979). However in talapoin monkeys and Japanese macaques there is no significant relationship between aggressive behavior and T (Eberhart et al., 1980; Eaton and Resko, 1974). These conflicting findings may result from differences in the number of groups examined and number of samples obtained. In several investigations, utilizing independent groups, the dominant male in each group did not always have higher T concentrations than all subordinate males (c.f. Eberhart et al., 1980). Several studies have relied on a single sample time to assess the relationship between T and aggression or dominance or have utilized only a single established group (Rose et al., 1971; Eaton and Resko, 1974). In view of the pulsate secretion of T and extreme short-term fluctuations in plasma T concentration (Plant, 1981; Castracane, Kyle, Wright, and Martinez, 1981), this approach may produce spurious behavioral-hormonal correlations. The present study addressed two related questions. One concerned the relationship between dominance rank and circulating concentrations of T in stable, captive, heterosexual groups of vervet monkeys (Cercopithecus aethiops sabaeus). Do dominant males have higher T titers than subordinate males, and, if so, is this difference consistent across sample times and groups? For this purpose, 15 social groups were examined and several blood samples were obtained from each male. The second question addressed the relationship between the frequency of aggressive interactions and T concentration. Do T titers correlate with the frequency of aggressive behavior, and, if so, which aspects of aggressive interactions (e.g., aggression initiated or received) are significantly related to T? These questions were explored in 4 groups where daily behavioral observation was followed by blood sampling. METHODS Subjects. Subjects were 20 adult male vervet monkeys (C. a. sabaeus) that were feral reared on St. Kitts and Nevis. These males were part of a total of 15 captive groups (see Table 1). Each group contained 2 or 3 adult males, 2 to 4 adult females, and their immature offspring. Eight groups were located at the St. Kitts Biomedical Research Foundation

156

STEKLIS

ET AL.

research site on St. Kitts, East Caribbean, while the remaining seven groups were housed at the Nonhuman Primate Laboratory at the Sepulveda Veterans Administration Medical Center, Sepulveda, California. All groups were housed in outdoor enclosures measuring at least 3 x 2 x 2 m. Enclosures contained perches, visual barriers, and areas sheltered from the sun and rain. All groups were maintained on isoniazid-free commercial monkey chow supplemented weekly with local produce. Groups had been established at least 10 weeks prior to the collection of hormonal and behavioral data. Behavioral data. Behavioral units were based on Struhsaker’s (1967) ethogram for this species and have been described in detail elsewhere (Raleigh, Brammer, Yuwiler, Flannery, McGuire, and Geller, 1980;Raleigh, Brammer, and McGuire, 1983). Observations were conducted between 0600 and 0930 and 1400 and 1800, the times when animals are most active. All occurrences of dyadic intermale aggressive behavior (displaying, threatening, contact aggressing, avoiding, and submitting) were recorded (Altmann, 1974). The initiator, recipient and outcome of each aggressive encounter were recorded. Dominance relationships were determined by noting an animal’s success in intermale aggressive encounters. A male was successful in an encounter if his opponent avoided or submitted to him. In each group the dominant male had the highest percentage of success. Over all 15 groups, the percentage of success ranged from 78 to 100for dominant males, compared to a range of 0 to 49 for subordinate males. Differences in rank between subordinate males are less apparent and do not persist as long as those between the dominant male and any subordinate male. Consequently, we made a binary distinction between dominant and subordinate males. No attempt was made to rank order the subordinate males. In four Sepulveda groups that were randomly selected at the outset of this study the association between aggression and T was assessed in 10 adult males. These four groups were selected without prior knowledge of the males’ testosterone levels. In these groups frequencies of contact aggression, display, and threat initiated and received by individuals, obtained from 3.5 hr (0600-0930) of observer contact per day over 4 consecutive days, were converted to rates/hour. Total rates of aggressive encounters won, lost, and percentage of success were also calculated. Blood sampling. All blood samples were obtained within 10 min of capture while subjects were under ketamine (10 mg/kg IM) anesthesia. Subjects were restrained in squeezecagesattached to their home enclosures and given ketamine, and blood was obtained by femoral venipuncture. Two-milliliter blood samples were collected into sterile, empty tubes at the same time of day from all males within a particular group. Blood was allowed to clot at ambient temperature and serum was obtained by centrifugation and frozen and stored at - 70°C. In 11 groups, blood was

SERUM

TESTOSTERONE

AND AGGRESSION

157

collected at 0800 on days for which no behavioral observations were scheduled. Four to six such samples were taken from each adult male at l- to 7-day intervals. Finally, in 4 groups nine samples were collected at 1200 at l- to 3-day intervals. Four of the nine samples were collected at 1200 after AM behavior data collection. Testosterone assay. Serum T concentration was determined by using acommercially available RIA kit (Diagnostic Products Corp., Los Angeles, Cahf.), which has a cross-reactivity of 22% with dihydrotestosterone and less than 5% with other androgens. All samples were assayed in duplicate, yielding an average intraassay coefficient of variation of 5.6%. Samples from an extract pool and three serum pools were included in each assay, from which an average interassay coefficient of variation of 11% (n = 38) was determined. Minimum assay sensitivity was 1000 pg/ml. RESULTS Table 1 shows that an overall striking finding was the marked withinsubject variation in T concentration. This was often observed to be 5- to IO-fold on successive days. Morning samples (0800) were no less variable than were noon samples, and there were no differences in variability of samples obtained from the Sepulveda groups compared to the St. Kitts groups. Hence for purposes of further analysis data from the St. Kitts and Sepulveda groups were combined. Figure 1 shows the mean T concentrations for dominant and subordinate males. Although overall dominant males had higher T concentrations (9879 + 1249 SEM, pg/ml, n = 15) than subordinate males (7931 2 777 SEM, n = 22), this difference was not significant (r(14) = 0.94, 0.25 < P < 0.10). Furthermore, differences between dominant and subordinate males in mean T concentrations were not significant in either three-male (r(6) = 1.36, NS) or two-male (t(7) = 0.01, NS) groups. Average T concentrations were higher in dominant than in subordinate males in 10 of 15 groups (see Table 1). When all days from which samples were available from each male were considered separately, dominant males had the highest absolute T concentration within their groups in 57 out of 95 total sample days. The marked within-subject variation in serum T concentration and the lack of a significant overall relationship between dominance and T suggests that serum T concentrations are related to fluctuations in daily rates of aggressive behavior within groups. To test this, behavioral and serum samples from four successive days were used from four gro~~ps, &puJve& groups 4-7, for a total of 16 samples for dominant and 24 samples for subordinate males. The behavioral measures (aggression initiated, aggression received, aggressive encounters won, aggressive encounters lost, percentage of successful or unsuccessful outcomes, and total aggressive encounters

158

STEKLIS ET AL.

TABLE I Serum Testosterone Concentrations (pg/ml) in Dominant (*) and Subordinate Males Groups” Sepulveda 1 2

Animal SS M G' F

3

E* K R F* B R’ K D’ A

7

t* G S

St. Kitts 1 2 3 4

6 7 8

W* X H* Y N* P C’ I Z P’ Y W Z* N V* H I Xf C

Mean concentration

Coefficient of variation

Number of samples

35 33 65 55 60 27 78 35 47 59 I4 53 38 50 56 47 72 97

4 4 6 6 6 6 6 6 9 9 9 9 9 9 9 9 9 9

43 48 87 47 91 48 53 54 58 66 48 91 56 62 51 I7 53 44 31

6 6 6 6 6 6 6 6 6 5 5 5 4 4 5 5 5 5 5

19720 6505 I0800 5055 1382 8679 12038 5311 8812 4335 6813 6750 7059 5049 2385 4921 4525 18620 8887 3914 16220 6017 8612 12523 10529 9310 11578 8691 7298 4188 14366 8891 15632 3116 11%2

’ Group numbers reflect temporal order of group formation.

SERUM TESTOSTERONE

Overall

(15)

AND AGGRESSION

Tim-Mok

Two-Mob

(7)

(61

159

FIG. 1. Mean serum testosterone concentrations in dominant and subordinate males in two- and three-male social groups.

engaged in) were examined for significant correlations (Pearson’s r, twotailed test) with serum T concentration. For all males, a significant correlation coeffcient was obtained between serum T concentration and rate of aggression initiated (r = 0.32, P < 0.05). As shown in Figure 2, separate analyses for dominant and subordinate males, however, revealed significant correlations only for dominant males between serum T concentration and rate/hour of aggression initiated (r = 0.57, P = 0.02) and total aggressive encounters engaged in (r = 0.55, P = 0.025). Similarly, when correlations between aggression and T concentration are examined

Aggressm

Imtioted

(Events

/hr)

2. Rate of aggression initiated as a function of serum testosterone concentrations in dominant and subordinate males. FIG.

160

STEKLIS ET AL.

separately for the two two-male and the two three-male Sepulveda groups, only for dominant males were there significant correlations (one-tailed test) in both types of groups between serum T titers and rates of aggression initiated (two- and three-male groups, respectively: r = 0.70, P = 0.025; r = 0.69, P < 0.05), aggressive encounters won (r = 0.69, P < 0.05; r = 0.62, P = 0.05). and total aggression (r = 0.71, P < 0.025; r = 0.74, P < 0.025). For all dominant males in these four groups, aggression initiated, encounters won, and total aggression were highly intercorrelated (initiated and won: r = 0.95, P < 0.001; initiated and total: r = 0.93, P < 0.001; won and total: r = 0.94, P < 0.001). As dominant males in these four groups won nearly all of the aggressive encounters engaged in (mean percentage: 91.5 2 3.3 SEM), aggression initiated and encounters won are not independently related to T titer. There was a trend, however, for aggression initiated to be more closely related to T concentration than rate of aggressive encounters won, as indicated by a partial correlational analysis involving these three variables (rate initiated and T: r = 0.41, f(13) = 1.63, NS; rate won and T: r = -0.23, ~(13) = -0.86, NS). The finding of a positive relationship between daily rate of aggression initiated and serum T concentration in these four groups led us to conduct a retrospective analysis of the relationship between mean aggression rate and mean T concentration for individual animals in all I5 groups. As all groups from which blood samples were obtained had also been observed regularly, employing methods identical to those described earlier (see Methods), it was possible to determine mean rate per hour of aggression initiated for all males. These means were based on periods of behavioral sampling that correspond as nearly as possible to the periods of blood sampling, although in the majority of cases behavioral data were not obtained on days of blood sampling. Means reflect 96, 14, 156, 12, and 32 hr of observer contact per group for Sepulveda groups l-3 and 4-7, and St. Kitts groups l-4, 5-6, and 7-8, respectively. When males of all 15 groups are considered together, the correlation between mean T concentration and mean rate of aggression initiated was marginally significant (r(35) = 0.30, P < 0.05, one-tailed test). In addition, in the 10 groups where dominant males had higher mean T concentrations than subordinate males, dominant males also had higher mean rates of aggression initiated (~(23) = 4.04, P < 0.001). Conversely, in the 5 remaining groups, where mean T concentration of dominant males did not exceed that of subordinate mates, there were also no differences between the two in mean rates of aggression initiated (t(l0) = -0.68, NS). These findings further indicate that differences between dominant and subordinate males in T titer are primarily related to differences in rate of aggression rather than dominance rank per se.

SERUM

TESTOSTERONE

AND AGGRESSION

161

DISCUSSION The present study indicates that in male vet-vet monkeys serum testosterone concentration is highly variable within subjects, unrelated to dominance rank per se, but positively associated with aggressive behavior in dominant males. The source of the extreme intrasubject variability in T concentration is currently unknown. Neither diurnal rhythms, seasonal factors, nor differences in social rank account for the intraindividual variability. Vet-vet and other Old World monkeys exhibit diurnal fluctuations in serum androgens (Michael, Setchell, and Plant, 1974; Beattie and Bullock, 1978). However, in this study all sets of successive samples from a given animal were obtained at the same time of day. Furthermore, intrasubject variability did not differ between samples obtained at 0800 and at 1200. For many species, T concentrations vary between breeding and birth seasons (Coe et al., 1983). However, in the present study intrasubject day-to-day fluctuations of up to IO-fold were observed. Furthermore at Sepulveda births are not limited to a particular season and all samples from St. Kitts groups were obtained during breeding season. Social status did not influence intrasubject variability in T: dominant and subordinate males did not differ in the amount of intrasubject variability in T. Among captive vervet monkeys, male dominance relationships are persistent and consistent over long periods of time. Once a group was established during the present study, no animal changed status (from dominant to subordinate or vice versa). Thus it is not surprising that a highly fluctuating biological parameter like serum T is unrelated to dominance status in vervet monkeys. The discrepancy between this observation and findings in other species may reflect species differences. However, it is also possible that this contrast results from other investigators’ reliance on T concentrations determined at a single point in time. The observation that when dominant and subordinate animals are considered together, T concentrations correlated (albeit weakly) with daily rates of aggressive behavior initiated supports the view that in primates T is associated with aggressivity rather than dominance rank per se. Since in this analysis aggression frequency was derived by combining frequencies for subcomponents (Le., display, threat, contact aggression), it is possible that T concentrations are more closely related to one or more of the subcomponents than to all of them combined. While this possibility merits further study, it cannot be addressed with the current data, because frequencies for subcomponents of aggression were too low to permit meaningful correlational analysis. It is also possible that sexual behavior additionally influences T in vervet monkeys. However, sexual behavior is very infrequent in captive vervet monkey groups. Indeed in the present study no instances of cop-

162

STEKLIS ET AL.

ulation were observed. Consequently, the relationship between T concentration and copulatory behavior could not be investigated. The strong positive association between aggression initiated and T in dominant but not subordinate males suggeststhat, while dominance status per se is unrelated to T titer, occupation of the dominant position does affect endocrine-behavioral relationships. In this and other species, biochemical and pharmacological studies underscore the impact of social status on physiological processes. For example, among free-ranging baboons (Sapolsky, 1982; 1983) there were status-linked differences in the effects of capture stress on T. Captive dominant male squirrel monkeys differ from subordinate squirrel monkeys in both basal and challenged testosterone and cortisol (Coe et al., 1983). There may also be statusrelated differences in the effects of fighting on T in rhesus monkeys (e.g., Perachio, 1978). The physiological basis for these status-related differences in endocrine-behavioral relationships are at present unknown. ACKNOWLEDGMENTS This research was supported in part by the Research Service of the Veterans Administration Medical CenterSepulveda and Brentwocd, in part by Giles and Elsie Mead Foundation, and in part by the Harry Frank Guggenheim Foundation. Helpful advice and guidance was provided by A. Yuwiler, J. Hershman, and E. Pekary. Technical assistance was provided by Melody Denham, Dave Flannery, Mary Hanington, Campbell Hughes, Sandra Lehman, Nuria Kimble, Jackie Nesbit, and Lewis Williams. The cooperation of the government and people of St. Kitts-Nevis is greatly appreciated. Jeffrey Flannery and Aileen Toshiyuki prepared the manuscript.

REFERENCES Altmann, J. (1974). Observational study of behavior: Sampling methods. Be/t&our 49, 227-267. Beattie, C. W., and Bullock, B. C. (1978). Diurnal variation of serum androgen and estradiol-l7@ in the adult male green monkey (Cercopirhecus sp.). Biol. Reprod. 19, 36-39. Bernstein, I. S., Rose, R. M., Gordon, T. P., and Grady, C. L. (1979). Agonistic rank, aggression, social context, and testosterone in male pigtail monkeys. Aggress. Behav. 5, 329-339. Bouissou. M. F. (1983). Androgens, aggressive behavior, and social relationships in higher mammals. Hormone Res. 18, 43-61. Castracane, V. D., Kyle, J., Wright, E., and Martinez, D. (1981). Episodic, circadian, and circannual patterns of plasma testosterone in the male baboon (Pupio cynocephalus). Amer. J. Primatol. 1, 344. Coe, C. L., Smith. E. R., Mendoza, S. P., and Levine, S. (1983). Varying influence of social status on hormone levels in male squirrel monkeys. In H. D. Steklis and A. S. Kling (Eds.), Hormones, Drugs, and Social Behavior in Primates, pp. 7-32. SP Medical and Scientific Books, New York. Dixson, A. F. (1980). Androgens and aggressive behavior in primates: A review. Aggress. Eehav. 6, 37-67.

Eaton, C. G., and Resko, J. A. (1974). Plasma testosterone and male dominance in a Japanese macaque (Mucaca fuscara) troop compared with repeated measures of testosterone in laboratory males. Harm. Behav. 5, 251-259.

163

SERUM TESTOSTERONE AND AGGRESSION

Eberhart, J. A., Keveme. E. B., and Meller, R. E. (1980). Social influences on plasma testosterone levels in male talapoin monkeys. Horm. Behav. 14, 247-266. Gordon, T. P., Rose, R. M., and Bernstein, 1. S. (1976). Seasonal rhythm in plasma testosterone levels in the rhesus monkey (Macaco mufaffa): A three year study. Horm. Behav. 7, 229-243.

Kling, A., and Dunne, K. (1976). Social-environmental factors affecting behavior and plasma testosterone in normal and amygdala lesioned M. speciosu. Primates 17, 23-42. Michael, R. P., Setchell, K. D. R., and Plant, T. M. (1974). Diurnal changes in plasma testosterone and studies on plasma corticosteroids in nonanaesthetized male rhesus monkeys (Macucu mulutra). J. Endocr. 63, 325-335. Perachio, A. A. (1978). Hypothalamic regulation of behavioral and hormonal aspects of aggression and sexual performance. In D. J. Chivers and J. Herbert (Eds.), Recent Advances in Primurology, Vol. 1, Behuviour, pp. 549-565. Academic Press, New York. Plant, T. M. (1981). Time courses of concentrations of circulating gonadotropin, prolactin, testosterone, and cortisol in adult male rhesus monkeys (Mucuca mulurra) throughout the 24 h light-dark cycle. Biol. Reprod. 25, 244-252. Raleigh, M. J., Brammer, G. L., Yuwiler, A., Flannery. J. W., McGuire, M. T., and Geller, E. (1980). Serotonergic influences on the social behavior of vet-vet monkeys (Cercopithecus

uethiops sabaeus).

Exp. Neural.

68, 322-334.

Raleigh, M. J., Brammer, G. L., and McGuire, M. T. (1983). Male dominance, serotonergic systems, and the behavioral and physiological effects of drugs in vervet monkeys (Cercopithecus uethiops subueus). In K. A. Miczck (Ed.), Ethophurmucology: Primate Models of Neuropsychiurric Disorders, pp. 185-197. A. R. Liss, New York. Rose, R. M., Holaday, J. W., and Bernstein, I. S. (1971). Plasma testosterone, dominance rank and aggressive behavior in male rhesus monkeys. Nature (London) 231, 3&i-368.

Rose, R. M.. Bernstein, I. S., and Gordon, T. P. (1975). Consequences of social conflict on plasma testosterone levels in rhesus monkeys. Psychosom. Med. 37, W-61. Sapolsky, R. M. (1982). The endocrine stress-response and social status in wild baboon. Harm. Behav.

16, 279-292.

Sapolsky, R. M. (1982). Endocrine aspects of social instability in the olive baboon (Pupio unubis).

Amer. J. Primutol.

5, 365-379.

Struhsaker, T. T. (1967). Behavior of vervet monkeys (Cercopithecus CaliS. Publ. 2001. 82, l-74.

aethiops).

Univ.