ILAR Journal Online, Volume 39(1) 1998: Animal Well-being: Immune Function, Behavior, Morphology, and Psychoneuroimmunology

Online Issues

<< All Back-issues

<< This Issue's Table of Contents

ILAR Journal V39(1) 1998
Animal Well-being: Immune Function, Behavior, Morphology, and Psychoneuroimmunology

Social Experience and Immune System Measures in Laboratory-housed Macaques: Implications for Management and Research
John P. Capitanio

John P. Capitanio, Ph.D., is Assistant Professor in the Department of Psychology and a Core Scientist at the California Regional Primate Research Center, University of California, Davis, California.

Since the revision of the Animal Welfare Act in 1985, the social needs of nonhuman primates living in the laboratory have received continuing attention. The general idea has been that because nonhuman primates are social creatures, their well-being is facilitated by providing social experience, either directly (through social housing, the preferred option) or indirectly (by placement of cages of individually housed animals within sensory (and especially visual) contact with each other) (de Waal 1991). The assessment of an animal's well-being has also been the subject of discussion (Novak and Suomi 1988).

In this review, I propose that (1) immune system-related measures are particularly useful indicators of an animal's internal state, (2) social experience of macaques should be viewed more broadly than is currently the norm, (3) research results may suffer when social experience is not so viewed, and (4) individual differences with respect to social experience are important to consider. Some of these ideas are not new. Coe, in particular, has previously raised the issue of whether social housing of primates is always the optimal choice (Coe 1991), also suggesting that immune measures may have utility in studying well-being (Coe and Scheffler 1989).

My goal is to place the discussion within a theoretical framework that can help organize the research results to be described and to describe data from psychoneuroimmunological research relevant to the issue of social experience in nonhuman primates. It is worth noting at the outset that many of the data to be presented were not obtained specifically to address issues of psychological well-being in captive primates, but rather were the results of studies focusing on basic psychobiological processes. Furthermore, my review focuses principally on data derived from research on the macaques, in general, and rhesus monkeys, in particular. The rationale for this choice is that the most extensive body of research exists for the macaques. Although many of the results described in this review are likely to be applicable to other primate species, the reader is cautioned that species differences do exist in some psychosocial processes that may be reflected in immune measures.

IMMUNE-RELATED MEASURES

The immune system comprises a diverse set of tissues whose principal function is to protect the organism from attack by potentially harmful microbes and their toxins. The immune system is complex and, as might be expected of a system with such an important task, is characterized by multiple, redundant subsystems that work both independently and in concert to achieve its effects. Importantly, for the present purposes, the immune system is in constant, bidirectional communication with other major physiological systems, including the endocrine and central nervous systems (see Ader's "Psychoneuroimmunology" in this issue).

Immune system activity can be measured in many ways, which generally fall into 2 categories--functional and enumerative (Cohen and Herbert 1996). Functional measures of immunity reflect the coordination of the various immune cells into a response to challenge with a pathogen (such as a disease outcome), an antigen (such as an in vivo measure of antibody production), or a mitogen (such as an in vitro proliferation assay of cellular immunity). Enumerative measures are those that reflect counts of various cell types, such as neutrophils, total lymphocytes, or lymphocyte subsets, such as CD4+ (helper) and CD8+ (cytolytic) T-cells. Functional measures provide the best indicators of how effectively the immune system responds to challenge. Typically, however, they take time to develop (the immunoglobulin G response to a novel antigen may take a week or 2 to develop) and so may reflect better the consequences of persisting, rather than acute, influences on immunity. Moreover, functional measures require some degree of expertise (and expense) to perform the requisite assays. Enumerative measures, however, are relatively simple to obtain--by counting cells either manually, on an electronic cell counter, or with a flow cytometer. The major disadvantage of enumerative measures is their unclear relationship to immune function. We still do not know whether a reduction in the number of lymphocytes in peripheral blood in response to some manipulation relates to how that animal might respond to a challenge by a pathogen or antigen. Evidence suggests the lack of a consistent relationship (Westermann and Pabst 1990).

Why should we consider enumerative measures if they do not necessarily interrelate directly with functional measures? The immune system is typically viewed as an effector organ that reacts to challenge. However, the fact that bidirectional communication exists between the immune and other physiological systems suggests another view: that the immune system is a sensory organ that "forms part of an integrated homeostatic network" (Husband 1995, p 377). Because the immune system plays an important role in keeping the animal alive, it makes sense for the immune system to be finely tuned to perturbations in the organism's internal state. These perturbations can arise from a number of sources, including infection and stress, and can be reflected in changes in operation of a number of physiological systems, any one (or all) of which affect the immune system. Moreover, activity in the immune system can have an effect on other systems of the body, including neuroendocrine and central nervous systems. This view, then, places the immune system within the larger physiological milieu that is affected by, and affects, those other systems. Thus, for our purposes, a social manipulation that causes a change in the physiological organization of the animal may not be detected if our outcome measure is simply plasma concentrations of cortisol. It will be detected, however, if we examine measures that are affected by direct innervation from the nervous system, by activation of adrenergic receptors, and by activation from pituitary-adrenal axis hormones. Leukocyte trafficking is affected by all of these routes, and enumerative measures of immunity reflect the convergence of these pathways (Ottaway and Husband 1992, 1994). In this sense, the immune system, integrated as it is with other systems of the organism, detects and reflects changes in the internal milieu resulting from multiple sources, including infection, and socioemotional processes.

Use of enumerative immune measures as indicators of the animal's internal status is not without problems, however. We have shown that obtaining such measures requires careful attention to blood sampling procedures (Capitanio and others 1996). For example, when we enter a room to draw blood sequentially on 9 conscious rhesus monkeys, we are sure that by the time we get to the ninth animal in the series, concentrations of plasma cortisol have not been affected by the general disturbance in the room associated with drawing blood on the preceding 8 animals. Rather, what affects the cortisol concentration for that ninth monkey is the duration of sample collection for that specific sample. Cortisol concentrations for samples obtained within about 3 min do not appear to reflect the blood drawing procedure. However, leukocyte numbers, which are responsive to so many inputs, change quite readily in response to the disturbance produced by other activity in the room. For example, 6 to 9 min of such disturbance can cause a 60% increase in the number of CD8+ T-cells in peripheral blood (Capitanio and others 1996). This suggests that use of immune measures to assess the effectiveness of a social manipulation requires careful control of blood sampling procedures and that disturbances before the blood sampling must be controlled, or at least equated, in the pre- and postmanipulation conditions, or for the experimental and control groups.

Peripheral blood hematological measures are also characterized by consistent individual differences. We sampled animals over a period of 13 mo under basal conditions (while in their individual living cages) and under conditions of acute physical restraint with and without pretreatment with dexamethasone (a potent synthetic glucocorticoid). We found significant rank-order correlations for a number of measures, including neutrophil and total lymphocyte numbers, and most especially the CD4:CD8 ratio (Capitanio and others 1998a). In other words, regardless of the effect of our manipulations on mean values across all animals (for example, 2 hr of physical restraint resulted in a tripling of the numbers of neutrophils in peripheral blood), those animals with relatively low counts at one time tended to have relatively low counts at other times. Consequently, a single, isolated result obtained from a blood sample drawn after a social manipulation may be difficult to interpret. Although high neutrophil numbers may reflect that the animal is stressed, it may also reflect characteristics of the specific individual's physiological organization.

A final methodological issue pertains to the timing of blood sampling. Again, because enumerative measures of immunity are influenced by many physiological systems, any circadian variation in systems that impinge on leukocyte trafficking will likely result in circadian variation in cell counts. Studies conducted by Abo's group have demonstrated circadian variation in leukocyte and lymphocyte subset distributions in humans that correlate with circadian variations in plasma concentrations of cortisol and catecholamines. CD4+ cells, for example, ranged from approximately 35% of T-cells at 1200 hr to nearly 45% of T-cells 16 hr later (Abo and others 1981; Suzuki and others 1997). With respect to functional measures, Hiemke and others (1995) found a 38% decrease between 0600 and 1000 hr in the proliferative response of T-cells to stimulation by tetanus toxoid. The timing of this decrease in proliferation corresponded to an increase in the plasma concentrations of cortisol resulting from the circadian rhythm. Attention to circadian effects can be accomplished most directly by drawing blood samples at the same time of day on all animals in the cohort.

To summarize, functional and enumerative measures of immunity provide slightly different information about the status of the organism. Functional measures can reflect more long-term influences on the organization of the individual. Enumerative measures are easily obtained, can reflect acute as well as persisting influences on functioning, and are valuable as sensitive indicators of homeostatic functioning. They can be informative about socioemotional changes resulting from an animal's change in social experience. They are not simple measures, however, and careful attention must be paid to the circumstances under which such data are obtained.

ORGANISMS AS SYSTEMS

An individual animal is a living system. Like all systems (such as those at the levels of cells, organ systems [like the immune system], social group, and culture), it is organized; it is composed of subsystems and is itself a component of other systems; and it is in dynamic exchange with its environment (Feibleman 1954; von Bertalanffy 1968). The organization of the individual at any given time is a function of its history (past transactions with its environment that result in individual characteristics seen later) and its current circumstances (inputs available to the individual, its appraisal processes, and the options available for action). General systems theory proposes that perturbations in a system reverberate at all levels of the system. Such a view is implied in our use of measures derived from one level (such as immunological or neuroendocrine) to assess changes we have imposed on the animal at another level (such as social).

A virtually infinite number of things can disturb a system. A disturbance may result in the activation of mechanisms that attempt to return the system to a homeostatic equilibrium. When the disturbance ends, the system may shift back to its original condition. If a disturbance persists, however, the system itself may undergo substantial change and possibly reorganize to accommodate the disturbance as well as it can. In such cases, relationships among the subsystems, set-points, thresholds of responsiveness, and so forth may change. If the disturbance occurs during a sensitive period (a time in development when a subsystem is developing most rapidly), the organism's developmental trajectory may be altered in a relatively permanent fashion. For example, a rhesus monkey reared without access to conspecifics for the first 6 mo of life shows very different behavior from a monkey reared with conspecifics during that period but without conspecifics for its second 6 mo of life (in Capitanio 1986). This result reflects the importance for the developing macaque of the establishment of a social relationship during the first few months of life.

SOCIAL FACTORS REGULATE PHYSIOLOGICAL ORGANIZATION

Many lines of evidence indicate that processes occurring at the social level of organization affect processes at lower levels, such as the individual organ (for example, immune) system, and tissue levels. Some very brief social-level events result in immunological events that persist for years, and other events produce more transitory consequences. The data reviewed below demonstrate the effects of rearing, separation, relocation, and group formation on integrated homeostasis, as indicated by immune-related measures.

Rearing Effects

The first indication that the rearing environment could have a lasting impact on immune functioning in macaques came from a study by Laudenslager and others (1985). Four pigtailed macaques that had experienced a single 10- to 14-day period of separation from mother and/or peers during the first year of life were contrasted to 5 animals that had received comparable experiences in the laboratory environment but had not undergone social separation in the first year. At the time of assessment, the animals were 4 to 7 yr old and had been living continuously in stable mixed-rearing-condition social groups. In vitro assessment of lymphocyte proliferation responses to mitogen stimulation revealed that the previously separated animals exhibited significantly reduced proliferation (by about half) to all 3 mitogens tested compared with the control animals. The previously separated animals also had 40% higher total leukocyte numbers in peripheral blood compared with controls. Subsequent research by Coe and colleagues (1989) confirmed the result of reduced proliferative responses for 2 of the 3 mitogens we used, with rhesus monkeys who had undergone repeated social separation between 3 and 7 mo of age but who had been socially housed in undisturbed conditions for nearly a year since that time.

Additional evidence of the long-term effects of rearing on immune function was provided in a series of studies contrasting mother-reared (MR¹)animals with nursery-reared (NR¹)animals that had received peer experience 3 times per week. Blood samples were drawn when the animals were 13 to 16 mo of age. Surprisingly, NR animals had enhanced immunological responses, as assessed by in vitro proliferative responses to mitogen stimulation; and the greater response to 1 mitogen, Con A, persisted when measured again 1.5 yr after the first assessment (Coe and others 1989). Interestingly, animals weaned from their mothers at 6 mo of age (rather than the usual 12 mo of age) displayed patterns of proliferation responses that were intermediate between the MR and NR monkeys. A subsequent longitudinal study by this group that contrasted other NR and MR monkeys confirmed that NR monkeys had an elevated proliferative response to mitogens, lower natural killer cell activity in vitro, and approximately 50% greater CD4:CD8 ratios than MR monkeys due primarily to a reduced percentage of CD8 cells in the peripheral blood of NR animals (Lubach and others 1995). It is important to note that in both studies, plasma cortisol concentrations were not different between rearing groups. Moreover, evidence presented in the second study suggests that the enhanced proliferative response did not correlate with better health: NR monkeys had a greater incidence of gastrointestinal difficulties requiring treatment compared with MR monkeys.

Together, these studies indicate that rearing conditions that vary from the species-typical norm of prolonged contact between mother and infant result in long-lasting changes in immune system activity. The apparently permanent nature of these changes suggests that important characteristics of the organization of the individual have been altered by the rearing condition. It is worth reiterating that the immune system measures were more sensitive to the different rearing conditions than was plasma cortisol, the more common physiological measure that might be used to detect such differences.

Effects of Separation from Mother

Some of the earliest research demonstrating the effects of social factors on physiological organization in macaques has come from studies of separation in laboratory-raised macaques (for example, Candland and Mason 1968). Early research focused on heart rate measures, showing elevations in infant heart rate and agitated behavior in the period immediately after removal of the mother. Often, within 24 hr, both behavioral and cardiac activity decreased to levels below preseparation baseline (Reite and Capitanio 1985). This research line was extended to include immunological measures, and results indicated that separation of an infant from its mother was associated with suppressed proliferative responses to mitogen stimulation (Laudenslager and others 1982). Subsequent research has focused on a number of factors that can influence immune responses to separations, such as the presence of familiar companions (which appears to buffer the infant: Laudenslager and Boccia 1996), familiarity with the separation environment (allowing the separated infant to remain in its home cage reduces the immune response to separation: Coe and others 1987), timing of the assessments (immunological effects of separation in juvenile squirrel monkeys are most apparent within a few days of separation: Friedman and others 1991), and individual differences (animals that respond to separation with greater behavioral distress appear to be most likely to show increased immunological disruption through the 2-wk separation period: Laudenslager and others 1990).

Relocation Effects

In a captive colony of animals, individuals may be relocated from 1 cage to another for a variety of reasons, including assignment to experimental protocols, reorganization of housing rooms, growth that necessitates a move to a larger cage, breeding purposes, establishment of groups, and so on. Although cage relocation has not been studied in great detail, there is evidence that relocation of an animal to a different cage, and especially to a different room, can result in elevated corticosteroid secretion, sleep disruption, and suppression of appetite and activity (Crockett and others 1993, 1995; Mason 1972; Mitchell and Gomber 1976; Phoenix and Chambers 1984). In fact, moving a macaque to a novel cage has been a common procedure used in experimental studies of stress responsiveness (Clarke and others 1994). There is also evidence that such experience can have persisting immunological consequences. A research program by Gust and colleagues has examined this issue most closely in several age/sex classes of rhesus macaques. Compared with control animals that remained in their natal group, 2-yr-old rhesus that were removed from their large, natal enclosures to individual housing for an 11-wk period showed significant elevations in plasma cortisol concentrations that persisted through week 8, significant reductions in total lymphocyte numbers (and especially the CD4+ subset) through the 11-wk period, and less weight gain. For example, at 11 wk after relocation, the number of CD4+ T-cells recorded for relocated animals was approximately 30% less than for the nonrelocated control animals. When returned to their natal social group, both experimental and control subjects showed elevated cortisol concentrations 24 hr later, but only the previously relocated subjects showed a 40 to 80% increase over baseline in the number of CD4+ and CD8+ lymphocytes over the 4-wk follow-up (Gordon and others 1992). A second study examined the role of having a preferred companion present during removal from a social group and subsequent housing in a novel cage for a 4-day period. Cortisol concentrations were elevated, and lymphocyte subset numbers were reduced compared with preremoval baseline, regardless of whether the companion was present. The reduction in lymphocyte subset numbers was, however, smaller for the females when they were housed with the preferred companion--at 96 hr, for example, the decrease from baseline in number of CD8+ cells was 17.5% in the companion condition but 53.8% in the alone condition. Taken as a whole, these data suggest that over the 4 days of the relocation, the measures for females with companions were returning more rapidly to preremoval baseline values (Gust and others 1994). In contrast to these results, removal of 4 adult male rhesus from a large enclosure to social housing with each other resulted in no significant changes in cortisol or lymphocyte subset numbers during the initial 24-hr period. When the animals were returned to their cage 1 yr later, however, a 66% increase in cortisol and a 30 to 35% decrease in lymphocyte subset numbers was found at the 24-hr time point. Not surprisingly, the animals showing the greatest changes were the ones that received the most severe aggression (Gust and others 1993). Together, these data suggest that removal from a social group, and the later return to the group, are events that have significant consequences for immune measures.

Are physiological changes resulting from relocations of sufficient magnitude that they can affect a disease process? Evidence we have collected suggests they are. We examined colony records of nearly 300 simian immunodeficiency virus (SIV¹)-inoculated rhesus monkeys from 4 regional primate research centers and determined the number of relocations they experienced (Capitanio and Lerche forthcoming). We defined a relocation simply as a change in housing from 1 cage to another, regardless of whether there were familiar, unfamiliar, or no companions left behind in the old cage or present in the new cage. After statistically controlling for variables likely to influence disease course (like viral strain), the number of relocations in the 90-day period preceding SIV inoculation significantly predicted survival, with animals that experienced more relocations showing reduced survival times. Moreover, experiencing a relocation sooner after viral inoculation, compared with later, was also associated with earlier mortality, suggesting that the disruption engendered by a relocation event might exert a disproportionate effect early after viral challenge.

We further classified the cage move history of the subjects based on whether they experienced a social separation. This was defined as the animal itself being relocated to a new cage, leaving its familiar companions behind, or the animal remaining in its own cage while familiar companions were moved out to other cages. The number of such separations experienced in the 90 days before inoculation was inversely related to survival, as was the frequency of separation in the 30 days after inoculation.

Many of the animals in our archive had been housed individually, especially after SIV inoculation. In what way could relocations influence the individually housed animal's physiological state and result in acceleration of their disease? It is likely that the mechanism involves activation of the stress response systems. Certainly there is an element of unfamiliarity associated with relocation to a novel room, or even a different cage in the same room. It is also likely, however, that relocations, even among individually housed animals, result in disruptions of social relationships and the need to establish new relationships with the animals housed opposite. Individually housed animals do interact with each other (Bryant and others 1988; Line and Morgan 1991; Schapiro and Bloomsmith 1995), and behaviors they display, such as threats, lipsmacks, and cage shakes, are used to establish and maintain social dominance relationships. Relocation, then, presents an animal with a novel situation, both physically and socially, which is likely to be perceived as stressful. The physiological data pertaining to relocations that were described at the beginning of this section are consistent with this interpretation. Thus, relocation events for individually housed animals, and separation events for group-housed animals, are events that are likely to have negative consequences that can persist for weeks. In the context of pathogenic challenge, the animal's health might well be compromised by such an event.

Group Formation

There is a long history of research investigating behavioral consequences of group formations in various species of Old World monkeys (for example, Bernstein and others 1974). More recently, 3 lines of investigation have focused on immune and disease consequences of these procedures. Adult female rhesus monkeys, all of whom had extensive social experience, but not with each other, were formed into a social group with a single adult male (Gust and others 1991). Observers recorded very little contact aggression, and the animals formed a dominance hierarchy within 48 hr. Cortisol was elevated 24 hr after the group formation, compared with preformation baseline, but not for the remainder of the 9 wk of the study. Similarly, numbers of CD4+ and CD8+ T-cells were approximately 40 to 50% lower than baseline at the 24-hr postformation time point and showed a return to baseline levels during the next 9 wk for the group as a whole. When dominance rank was considered, however, the 4 low-ranked females had suppressed T-cell subset numbers that persisted through the study period, compared with the 4 highranked females (Gust and others 1991). In a second study, juvenile rhesus monkeys were formed into a new social group. Some of the juveniles came from the same social group, and others came individually from different social groups and so did not know any of their new group-mates. All animals showed a decline in numbers of immune cells 24 hr after group formation. However, animals with companions showed a significantly smaller decrement in their immune cell numbers than animals who had no familiar companions in the new group. Cortisol showed a comparable pattern, with the "alone" subjects showing a greater change from baseline than the "companion" subjects (Gust and others 1996).

In our retrospective analysis of SIV-inoculated rhesus monkeys (Capitanio and Lerche forthcoming), we identified 31 animals that had been socially housed at some point during the first 30 days of postinoculation: 8 animals had been housed in groups of 8 to 12 animals as part of studies investigating horizontal transmission; 17 animals had been pair-housed for enrichment purposes; and 6 animals were females that had been living with recently born infants. Pair- and group-housed animals were combined for the analysis inasmuch as both sets of animals were predominantly male and had been housed with like-aged peers and all were younger than 4 yr. Social housing was significantly related to survival: Pair/group-housed animals (but not mother-infant pairs) survived for a shorter period than individually housed animals. In fact, of all psychosocial measures we examined in this retrospective study, social housing showed the strongest effect on the death rate, and it was negative. This result seemed counterintuitive until we examined the composition of the groups more closely. For all pair-housed monkeys, animals were paired with previously familiar animals before SIV inoculation. In most cases, however, the animals were paired very close to the day of inoculation and after a period of individual housing. Only in a few cases were animals removed from large social groups and pair-housed in indoor cages. Thus, in most cases, animals were required to reestablish social relationships with previously familiar animals in very different social contexts, after a period of individual housing, often just before inoculation. Among the group-housed animals, the social groups were formed using animals of mixed familiarity--some animals were familiar with each other and some were not, a procedure that is not the most effective for minimizing interanimal aggression. Moreover, the 2 groups were formed only 1.5 mo or 6 days before SIV inoculation. Clinical records indicated that 5 of the 8 animals experienced trauma within the first month after inoculation. These data suggest that the inoculations may have been initiated at a time when the animals' physiology had not stabilized from the social manipulations.

Finally, in a prospective, experimental study of the effect of social relationships on neuroendocrine function, immune activity, and disease progression in SIV-inoculated monkeys, we found that social instability was associated with significantly faster disease progression and shorter survival (Capitanio and others 1998b). All animals were individually housed and given social experience 100 min per day. Animals in the stable social condition were formed into 3-member groups, the composition of which did not change during the study. Animals in the unstable condition were formed into groups in which the number and identity of the partners changed daily. This daily disruption of social context for the animals inhibited the formation of stable social bonds among the animals. Median survival of the stable animals was 40% longer than for the unstable animals. In addition, the 2 groups showed an unexpected pattern of plasma cortisol concentrations: Within 8 wk of establishing the social conditions, basal levels of cortisol declined for all animals. Cortisol levels among stable animals subsequently rose, but cortisol continued to decline for unstable animals through 24 wk of postinoculation. Such a pattern of cortisol has been found in other primate species during group formation studies (Mendoza and others 1991) and appears to reflect long-term, socially induced alteration in regulation of the hypothalamic-pituitary-adrenal axis.

MECHANISMS UNDERLYING SOCIALLY INDUCED ALTERATION IN IMMUNE ACTIVITY

What might be the mechanisms by which social factors influence integrated homeostasis? A thorough treatment of this question is beyond the scope of this review. Several studies described in prior sections, however, have suggested that socially induced changes in hypothalamic-pituitary-adrenal activity, one of the organism's major stress-response systems, might be involved. Those studies demonstrated changes in cortisol concentrations that coincided to some extent with changes in immune cell numbers in peripheral blood. In addition, other research has suggested that some of the social manipulations described are likely to affect operation of the second major stress response system, the sympathetic-adrenal-medullary system. In fact, glucocorticoids have been found to influence production of cytokines from immune system cells (Munck and Guyre 1991 ), and lymphocytes have been found to contain beta-adrenergic receptors, with Maisel and others (1989) reporting that CD8+ cells have numbers greater than CD4+ cells.

Can we assume that socially induced alterations in operation (and perhaps regulation: Capitanio and others 1998b) of stress-response systems explain all of the immune changes described here? The answer is probably not: Leukocyte trafficking is a complex process (for example, Butcher 1990) involving regulation and activation of receptors, expression of and interaction among adhesion molecules on the surfaces of both immune and endothelial cells, and cytokine production and secretion by cells. Moreover, the psychosocial manipulations described involve more than just increased cortisol or norepinephrine output. Laudenslager and others (1995), for example, have found that growth hormone increases in pigtail, but not bonnet macaque, infants after separation from mother. Moreover, as Lubach and colleagues (1995) note, NR monkeys experience more than a change in the nature and number of companions compared with MR monkeys. For example, NR animals experience lower nocturnal body temperature, and differences in their diurnal temperature rhythm, compared with MR animals (Lubach and others 1992). Furthermore, NR infants do not receive the benefits of maternal breast milk, particularly the immunological components passed from mother to offspring, and they grow up often having experienced increased health problems and having consumed a substantial amount of heterospecifically derived (typically bovine) protein contained in formula (Lubach and others 1995). Consequently, it may not be surprising that events experienced by an individual during a period when it is undergoing rapid change may result in that system having somewhat different "operating characteristics" compared with individuals having experienced a different constellation of events at the same time.

IMPLICATIONS FOR MANAGEMENT AND RESEARCH

I must reiterate here that the data reviewed above were derived primarily from studies of rhesus macaques. There are important species differences among nonhuman primates (including those species most likely to be found in laboratory colonies) with respect to the importance of particular psychosocial processes, which reflect, at least in part, differences in social organization. Even among macaques, for example, differences in aspects of stress responsiveness (Clarke and others 1994) and response to maternal separation (Laudenslager and others 1990) have been found. Nevertheless, I believe the 4 points mentioned at the beginning of the article and discussed below apply broadly to laboratory-housed nonhuman primates.

Value of Immune System Measures

Measures of immune system activity can be regarded as the endpoints of multiple physiological processes. They provide sensitive indicators of integrated homeostasis and can serve to indicate whether a change has taken place in an animal's internal milieu. This is particularly true of enumerative measures of immunity. Such measures are easily obtained, although their sensitivity to context necessitates close attention to the conditions under which blood samples are drawn.

Viewing the Social Experience of Primates More Broadly

An individual animal is a complex biological system. Activity at all levels of organization influence the animal's functioning at the other levels. Viewed in this way, it is clear that even the individually housed animal exists in a context, the disruption of which will influence all levels. I believe the data presented above suggest that our view of the social experience of animals may not always correspond with the animals' own perceptions. Because animals housed opposite each other establish social relationships, relocation of individually housed animals should be considered from the perspective of social disruption in much the same way as relocation of animals living in social groups. Similarly, it is important to recognize that periodic provision of social experience to NR infants does not result in an animal that shows the same physiological organization as one reared in a social group. Both sets of animals have "social experience," but the outcomes are quite different. In addition, the data also suggest that animals left behind when others are removed (Laudenslager and Boccia 1996), or animals already present when previously removed animals are reintroduced into a group (Gordon and others 1992), also show physiological changes, even though other familiar companions may be present.

I believe the data also suggest that implementation of social housing to enhance well-being should receive more careful consideration. First, animals' relationships with each other have contextual components: Placing together 2 animals familiar with each other in an earlier context (such as a large social group) in a new and very different context (such as a small cage with no other companions) after they have been separated from each other may result in some difficulties as the animals readjust their relationships. In contrast, moving the animals simultaneously from 1 context to the other appears to serve as a buffer against the stressful consequences of relocation. Finally, continuous housing, particularly when an animal has little control over the social situation (that is, no place to escape from unwanted social interactions), may not be equally enriching to both members of a pair. Many aspects of social interaction should be considered beyond simple presence or absence of a companion. Psychoneuroimmunological and stress research have shown that control and predictability are important psychological factors for health and well-being in humans, and both are likely to be important also for nonhuman primate well-being.

Impact on Research Results

The data presented above indicate that a variety of social manipulations have an effect on immune outcome measures. Are the changes of sufficient magnitude to affect research results? The retrospective analysis of SIV-inoculated rhesus monkeys suggests that they can be significant. Furthermore, those data indicated that timing of social manipulations was a crucial consideration: Relocations that occurred closer in time to the inoculation appeared to have a disproportionately negative effect on outcome. One implication is that careful attention should be paid to the beginning time of an animal's participation in a project relative to when it experienced the last social disruption, especially if the project is concerned with studying disease processes. Our data suggest, for example, that stability in physiology has not yet occurred 1 mo after relocation from social housing to individual housing; in contrast, we found fewer differences between basal blood samples taken at 5 and 10.5 mo after relocation (Capitanio and others 1998a). Those data, combined with the relocation data described earlier, suggest that the physiological changes after such a social disruption may require a few months to become stable.

Importance of Individual Differences

Several of the studies reviewed above showed that the magnitude of the immune response was directly related to the magnitude of the behavioral response to a separation or reunion. In other words, animals who responded to these events with less distress showed a reduced response. Moreover, as indicated above, social housing may not be perceived as equally beneficial to both members of a pair; Gust and colleagues (1991) reported that immune consequences of group formation persisted for 9 wk for the subordinate animals only, a result that echoes earlier results by Coe (1991). In addition, individual differences exist in immune measures (Capitanio and others 1998a; Laudenslager and others 1996), and our own unpublished data indicate that differences between individuals in personality variables such as "sociability" may be related to how well individuals respond immunologically to social manipulations such as relocations to individual housing or exposure to socially unstable conditions. Thus, attempts to select individuals that are as homogeneous as possible for research projects should include not only the usual criterion of age/sex class, but also consideration of other measures that might have an impact on the data.

Conclusions

This review is not intended to suggest that every social manipulation of animals is unwarranted or detrimental. It is intended, however, to show that even the most trivial of manipulations may have physiological consequences, which can persist for years in some cases. Moreover, manipulations that were intended to ameliorate 1 problem (such as the abnormal behavior that results from lack of conspecific interaction leading to nursery-rearing, which incorporates periodic socialization), may result in other problems (such as the different physiological organization of MR versus NR individuals). In the end, a greater awareness and appreciation of the animals' social experiences and individual differences in characteristics may result in physically and psychologically healthier animals, reduced variability in our data, and stronger research results obtained, possibly, from fewer animals.

¹Abbreviations used in this article: MR, mother-reared; NR, nursery-reared; SIV, simian immunodeficiency virus.

ACKNOWLEDGMENTS

I thank S. P. Mendoza and the anonymous reviewers for providing helpful comments on an earlier draft of this manuscript. This research was supported by National Institutes of Health grants MH49033 and RR00165.

REFERENCES

Abo T, Kawate T, Itoh K, Kumagai K. 1981. Studies on the bioperiodicity of the immune response. I. Circadian rhythms of human T, B, and K cell traffic in the peripheral blood. J Immuno1 126:1360-1363.

Bernstein IS, Gordon TP, Rose RM. 1974. Factors influencing the expression of aggression during introductions to rhesus monkey groups. In: Holloway RL, editor. Primate Aggression, Territoriality, and Xenophobia. New York: Academic. p 211-240.

Bryant CE, Rupniak NMJ, Iversen SD. 1988. Effects of different environmental enrichment devices on cage stereotypies and autoaggression in captive cynomolgus monkeys. J Med Primatol 17:257-269.

Butcher EC. 1990. Cellular and molecular mechanisms that direct leukocyte traffic. Am J Path 136:3-11.

Candland DK, Mason WA. 1968. Infant monkey heart rate: Habituation and effects of social substitutes. Dev Psychobiol 1:254-256.

Capitanio JP. 1986. Behavioral pathology. In: Mitchell G, Erwin J, editors. Comparative Primate Biology. Vol 2A: Behavior, Conservation, and Ecology. New York: Alan R. Liss. p 411-454.

Capitanio JP, Lerche NW. Social separation, housing relocation, and survival in simian AIDS: A retrospective analysis. Psychosom Med (Forthcoming).

Capitanio JP, Mendoza SP, Lerche NW. 1998a. Individual differences in peripheral blood immunological and hormonal measures in adult male rhesus macaques (Macaca mulatta): Evidence for temporal and situational consistency. Am J Primatol 44:29-41.

Capitanio JP, Mendoza SP, Lerche NW, Mason WA. 1998b. Social stress results in altered glucocorticoid regulation and shorter survival in simian acquired immune deficiency syndrome. Proc Natl Acad Sci U S A 95: 4714-4719.

Capitanio JP, Mendoza SP, McChesney M. 1996. Influences of blood sampling procedures on basal hypothalamic-pituitary-adrenal hormone levels and leukocyte values in rhesus macaques (Macaca mulatta). J Med Primatol 25:26-33.

Clarke AS, Mason WA, Mendoza SP. 1994. Heart rate patterns under stress in three species of macaques. Am J Primatol 33:133-148.

Coe CL. 1991. Is social housing of primates always the optimal choice? In: Novak MA, Petto A J, editors. Through the Looking Glass: Issues of Psychological Well-being in Captive Nonhuman Primates. Washington DC: American Psychological Association. p 78-92.

Coe CL, Lubach GR, Ershler WB, Klopp RG. 1989. Influence of early mating on lymphocyte proliferation responses in juvenile rhesus monkeys. Br Behav Immunol 3:47-60.

Coe CL, Rosenberg LT, Fischer M, Levine S. 1987. Psychological factors capable of preventing the inhibition of antibody responses in separated infant monkeys. Child Dev 58:1420-1430.

Coe CL, Scheffler J. 1989. Utility of immune measures for evaluating psychological well-being in nonhuman pirates. Zoo Biol Suppl 1:89-99.

Cohen S, Herbert TB. 1996. Health psychology: Psychological factors and physical disease from the perspective of human psychoneuroimmunology. Ann Rev Psychol 47:113-142.

Crockett CM, Bowers CL, Sackett GP, Bowden DM. 1993. Urinary cortisol responses of longtailed macaques to five cage sizes, tethering, sedation, and room change. Am J Primatol 30:55-74.

Crockett CM, Bowers CL, Shimoji M, Leu M, Bowden DM, Sackett GP. 1995. Behavioral responses of longtailed macaques to different cage sizes and common laboratory experiences. J Comp Psychol 109:368-383.

de Waal FBM. 1991. The social nature of primates. In: Novak MA, Petto AJ, editors. Through the Looking Glass: Issues of Psychological Well-being in Captive Nonhuman Primates. Washington DC: American Psychological Association. p 69-77.

Feibleman JK. 1954. Theory of integrative levels. Br J Philos Sci 5:59-66. Friedman EM, Coe CL, Ershler WB. 1991. Time-dependent effects of peer separation on lymphocyte proliferation responses in juvenile squirrel monkeys. Dev Psychobiol 24:159-173.

Gordon TP, Gust DA, Wilson ME, Ahmed-Ansari A, Brodie AR, McClure HM. 1992. Social separation and reunion affects immune system in juvenile rhesus monkeys. Physiol Behav 51:467-472.

Gust DA, Gordon TP, Brodie AR, McClure HM. 1994. Effect of a preferred companion in modulating stress in adult female rhesus monkeys. Physiol Behav 55:681-684.

Gust DA, Gordon TP, Brodie AR, McClure HM. 1996. Effects of companions in modulating stress associated with new group formation in juvenile rhesus macaques. Physiol Behav 59:941-945.

Gust DA, Gordon TP, Hambright MK. 1993. Response to removal from and return to a social group in adult male rhesus monkeys. Physiol Behav 53:599-602.

Gust DA, Gordon TP, Wilson ME, Ahmed-Ansari A, Brodie AR, McClure HM. 1991. Formation of a new social group of unfamiliar female rhesus monkeys affects the immune and pituitary adrenocortical systems. Br Behav Immunol 5:296-307.

Hiemke C, Brunner R, Hammes E, Muller H, Meyer zum Buschenfelde K-H, Lohse AW. 1995. Circadian variations in antigen-specific proliferations of human T lymphocytes and correlation to cortisol production. Psychoneuroendocrinology 20:335-342.

Husband AJ. 1995. The immune system and integrated homeostasis. Immunol Cell Biol 73:377-382.

Laudenslager ML, Berger CL, Boccia ML, Reite ML. 1996. Natural cytotoxicity toward K562 cells by macaque lymphocytes from infancy through puberty: Effects of early social challenge. Br Behav Immunol 19:275-287.

Laudenslager ML, Boccia ML. 1996. Some observations on psychosocial stressors, immunity, and individual differences in nonhuman primates. Am J Primatol 39:205-221.

Laudenslager ML, Boccia ML, Berger CL, Gennaro-Ruggles MM, McFerran B, Reite ML. 1995. Total cortisol, free cortisol, and growth hormone associated with brief social separation experiences in young macaques. Dev Psychobiol 28:199-211.

Laudenslager ML, Capitanio JP, Reite ML. 1985. Possible effects of early separation experiences on subsequent immune function in adult macaque monkeys. Am J Psychiatry 142:862-864.

Laudenslager ML, Held PE, Boccia ML, Reite ML, Cohen JJ. 1990. Behavioral and immunological consequences of brief mother-infant separation: A species comparison. Dev Psychobiol 23:247-264.

Laudenslager ML, Reite ML, Harbeck R. 1982. Suppressed immune response in infant monkeys associated with maternal separation. Behav Neurol Biol 136:40-48.

Line SW, Morgan KN. 1991. The effects of two novel objects on the behavior of singly caged adult rhesus macaques. Lab Anim Sci 41:365-369.

Lubach GR, Kittrell EMW, Coe CL. 1992. Maternal influences on body temperature in the infant primate. Physiol Behav 51:987-994.

Lubach GR, Coe CL, Ershler WB. 1995. Effects of early rearing environment on immune responses of infant rhesus monkeys. Br Behav lmmunol 9:31-46.

Maisel AS, Fowler P, Rearden A, Motulsky HJ, Michel MC. 1989. A new method for isolation of human lymphocyte subsets reveals differential regulation of beta-adrenergic receptors by terbutaline treatment. Clin Pharmacol Ther 46:429-439.

Mason JW. 1972. Corticosteroid response to chair restraint in the monkey. Am J Physiol 222:1291 - 1294.

Mendoza SP, Lyons DM, Saltzman W. 1991. Sociophysiology of squirrel monkeys. Am J Primatol 23:37-54.

Mitchell GD, Gomber J. 1976. Moving laboratory rhesus monkeys (Macaca mulatta) to unfamiliar home cages. Primates 17:543-547.

Munck A, Guyre PM. 1991. Glucocorticoids and immune function. In: Ader R, Felten DL, Cohen N, editors. Psychoneuroimmunology. 2nd ed. Orlando: Academic Press. p 447-474.

Novak MA, Suomi SJ. 1988. Psychological well-being of primates in captivity. Am Psychol 43:765-773.

Ottaway CA, Husband AJ. 1992. Central nervous system influences on lymphocyte migration. Br Behav Immunol 6:97-116.

Ottaway CA, Husband AJ. 1994. The influence of neuroendocrine pathways on lymphocyte migration. Immunol Today 15:511-517.

Phoenix CH, Chambers KC. 1984. Sexual behavior and serum hormone levels in aging rhesus males: Effects of environmental change. Horm Behav 18:206-215.

Reite M, Capitanio JP. 1985. On the nature of social separation and social attachment. In: Reite M, Field T, editors. The Psychobiology of Attachment and Separation. Orlando: Academic Press. p 223-255.

Schapiro S J, Bloomsmith MA. 1995. Behavioral effects of enrichment on singly housed, yearling rhesus monkeys: An analysis including three enrichment conditions and a control group. Am J Primatol 35:89-101.

Suzuki S, Toyabe S, Moroda T, Tada T, Tsukuhara A, Hai T, Minagawa M, Maruyama S, Hatakeyama K, Endoh K, Abo T. 1997. Circadian rhythm of leucocytes and lymphocyte subsets and its possible correlation with the function of the autonomic nervous system. Clin Exp lmmunol 110:500-508.

von Bertalanffy L. 1968. General system theory--A critical review. In: Buckley W, editor. Modern Systems Research for the Behavioral Scientist. Chicago: Aldine. p 11-30.

Westermann J, Pabst R. 1990. Lymphocyte subsets in the blood: A diagnostic window on the lymphoid system? Immunol Today 11:406-410.