Main

In humans, linear growth depends on multiple factors, including genetic constitution, nutrition, systemic disease, hormonal balance, and psychosocial environment(1–6). Particularly critical for normal growth and behavioral development appears to be the interaction between the infant and its primary caregiver(s)(6). In this regard, the primary caregiver(s) not only provides sustenance by tending to the infant's physiologic needs, such as nutrition and maintenance of body temperature, but also acts as a source for emotional security(7, 8).

Disruption of the infant-caregiver relationship, as in maternal deprivation, rejection, or abuse, may contribute to marked physiologic and behavioral abnormalities in the offspring(9, 10). In particular, caregiver inadequacy has been associated with both impaired skeletal growth and an inability to form social bonds during development(11, 12). In children, psychosocial dwarfism is thought to result from the withdrawal of normal “care” by the parent(4) and has been characterized by: 1) both lower than normal body weight and height and delayed onset of puberty;2) retardation of intellectual development, including regressive or bizarre behaviors and retarded psychomotor development; and 3) delayed social maturation and psychosexual development(13, 14). In primates, active manipulation of the mother-infant relationship through isolation or separation also appears to be associated with physiologic changes and long-lasting behavioral effects in the offspring(15, 17). However, it should be noted that spontaneously occurring variation in primate mothering also appears able to influence infant behavior(18).

The common marmoset, Callithrix jacchus jacchus, is a small, social, New World primate. Common marmosets are generally described as monogamous and live in extended family groups(19). Infant care in these monkeys appears to be dependent upon the cooperation of the entire family(20). Marmosets reproduce rapidly and tend to have twins or triplets which exhibit rapid development(21). The young monkey reaches puberty between 50 to 60 wk and is considered an adult by about 2 y of age(22). These developmental features make the marmoset attractive for examining the role of psychosocial interaction in development. Furthermore, there is evidence which suggests that the infant common marmoset may be vulnerable to early disruptions in caregiver bonds(23).

We report here a study of the biobehavioral consequences of spontaneously occurring variation in early caregiving behaviors in marmosets. Body weight and indices of stature, head circumference, and genital size were assessed at six different time points from birth to puberty and were correlated with early caregiver-infant interactions. Behavior during development was also assessed to examine the possible effects of early care on subsequent social development. Finally, to examine the HPA axis function in the young adult, we administered oCRH and measured the plasma IR ACTH and IR cortisol response.

METHODS

Subjects. Subjects were 18 female and 8 male infant common marmosets born into 13 family groups. The original breeding pairs were captive born, bred from wild stock originating from South America. During this study, infants were left undisturbed in their natal family groups. The family group composition was not manipulated and, at maximum, consisted of a mated female-male pair and two to three sequential sets of offspring born at approximately 152-d intervals.

The colony was maintained according to federal guidelines for laboratory animal care and as previously described(24). Animals were housed under an artificial light/dark cycle of 12 h light: 12 h dark on time at 0600 h. The room was kept at a temperature of 23-28°C and a relative humidity of 50-60%. Food and water were available ad libitum.

Physical measurements. Physical measurements were taken on unanesthetized monkeys at 3, 6, 10, 20, 35, and 50 wk of age. Both parents and infants were captured by hand by one investigator, a procedure to which the animals were habituated through the use of banana pellets as treats. Measures of physical development and growth included body weight, knee-heel length, crown-tail length, and head circumference. Indicators of sexual development and maturation were provided by pudendal pad width in females and testis volume in males which were measured using calipers. The volume (V) of the right testis was calculated from measures of width (W) and length (L) of the right testis using the formula V =π1/6W2L.

Behavioral observation. Behavioral information on the care of the infant was gathered by focal animal sampling (continuous sampling)(25). Infants were observed in their home cage twice each week from birth to 50 wk of age. Two 15-min samples were taken between 0900 and 1300 h and the animals were observed in a randomly determined order. Observations were performed by 1 of 2 investigators (E.O.J.), both of whom performed the physical measures. The monkeys had become habituated to these investigators by extended daily exposure over the entire length of the study. Interobserver reliability was regularly assessed from data collected simultaneously. Within subject variations in frequencies and total durations of the behaviors recorded correlated highly between the two observers; in all cases the correlation coefficients were greater than 0.85.

Behavioral measurements of parenting concentrated on discrete, quantifiable behaviors directed by the caregiver toward the infant. The mother, father, and older siblings are referred to collectively as “caregivers.” The two categories of behaviors scored were: positive/caring and negative/punishment behaviors. The frequency of the following behaviors was recorded:

Positive/caring behaviors. 1) Carry: the caregiver supports all four limbs of the infant. Infant is usually on the back of the caregiver. Carrying bouts usually lasted for several minutes, thus we were able to assess the duration of carrying (percentage of observation time infant was carried) by using a stop-watch during each session. 2) Retrieval: retrieval or active attempt to retrieve infant from another animal or while it is independent. 3) Food sharing: caregiver feeds the infant solid food or allows the infant to take or eat food from its hand or mouth. 4) Huddle: the caregiver and infant sit stationary in physical contact.5) Groom: caregiver combs through infant's pelage with hand and teeth. 6) Social exploration: active manipulation by the caregiver of infant with intense olfactory and gustatory investigation of the infant's anogenital area. 7) Play: social play which may consist of wrestle, chase, pause, grasp, and pounce (any or all) as described by Chalmers and Locke-Haydon(26).

Negative/punitive behaviors. 1) Rub off: caregiver successfully or unsuccessfully terminates the carrying bout by rubbing the infant against part of the cage until the infant is removed. 2) Rejection: caregiver prevents or tries to prevent the infant from climbing onto its back and initiating a carrying episode. 3) Bite/grab: caregiver aggressively bites or yanks at the infant.

After investigating the distribution of behavioral frequencies over time, data were grouped and averaged over the total observation time into five periods representing different phases of marmoset development: 1) wk 1-3, representing the period of infancy, defined as the period of caregiver dependence; 2) wk 4-6, weaning; 3) wk 7-10, early independence; 4) wk 11-20, young juveniles; and 5) wk 21-50, preadolescence, spanning prepuberty to puberty. Negative, abusive behaviors occurred spontaneously in a few family groups. These behaviors consisted predominately of serious biting or nipping of the infants tail by both parents to the extent that the lower third to half of the tail was amputated. Older siblings were never observed to participate in these abusive behaviors. Abuse was observed only during the first 2-3 wk after birth. Infants that experienced this type of spontaneous aggression are referred to as “abused infants.”

To assess the total amount of positive care received by infants without weighing the relative importance of positive to negative behaviors, we derived a coefficient of care. This more global measure of “care,” or nurturing, was derived by calculating the sum frequency of all punitive behaviors subtracted from the sum frequency of all caring behaviors.

Statistical comparisons of behaviors were made using two-tailed nonparametric tests. Behavioral changes across time were analyzed first with Friedman's test (ANOVA) for related measures; the Mann-Whitney U test was used for posthoc comparison of behaviors between abused and normal care animals. Statistical comparisons of stature were made using two-tailed parametric tests (ANOVA followed by Fisher PLSD). Correlations between care and indices of stature were analyzed by the Spearman rank coefficient(ρ).

CRH stimulation test. Pituitary-adrenal responsiveness to i.v. oCRH administration was assessed during the normal plasma ACTH and cortisol nadir. oCRH (Bachem Inc., Torrance, CA) was diluted with sterile water and given i.v. at a dose of 3 μg/kg of body weight at 0800 h. Blood samples were drawn immediately before and at 60 and 180 min after oCRH injection for determination of plasma IR ACTH and IR cortisol concentrations.

Blood samples (0.30 mL) were obtained via femoral veni-puncture from unanesthetized, manually restrained monkeys using tuberculin syringes (1 mL) with a 27-gauge needle. Blood samples were collected into chilled tubes containing EDTA and placed immediately on ice. The process of collecting blood samples was complete within 1 min of opening the monkey's cage, and all animals had been habituated to the process. Samples were centrifuged at 2500 rpm for 20 min at 4°C, and the plasma was extracted and frozen at-70°C until assayed. Each monkey was given an iron supplement after every sampling.

A single RIA for each hormone was performed at the end of the study. Plasma for ACTH determinations was extracted on Sep-pak cartridges (Waters, Milford, MA) as previously reported(27). The mean recovery of ACTH added to hormone-free plasma by this extraction procedure was 95.1%. Plasma ACTH levels were measured by RIA using a anti-ACTH serum, purchased from IgG Corp. (Nashville, TN). This anti-ACTH serum was used at a final dilution of 1:21,000 (assay volume, 0.3 mL), which bound 34.7 ± 5.5% of the 125I-ACTH tracer. Nonspecific binding was 2.3 ± 0.2%. The detection limit (ED90) was 14.4 pg/mL of plasma, and the intraassay coefficient of variation was 2.8%. Plasma cortisol was measured directly, without prior extraction using a 125I-cortisol kit (Diagnostic Products Corp., Los Angeles, CA). Total and nonspecific binding were 69 and 0.9%, respectively. The ED90 was 0.2 μg/dL and the intraassay coefficient of variation was 3.05%.

RESULTS

Normal growth. Males and females in our study did not differ significantly in their patterns of body weight gain, knee-heel length, head-tail length, or head circumference increase from birth to 50 wk of age. Similarly, there were no significant differences in the adult body weight and size between males and females. Table 1 lists the mean(±SEM) measures of physical parameters for 20 adult females and 16 adult males who were not subjects of the present study, but were used as controls. The variation in measures among all animals of the same age was low at all ages. The pudendal pad in females was dark brown at birth, became light pink between 6 and 10 wk, and attained adult size (13.61 ± 0.59 mm) by 50 wk of age. The testes descended into the scrotum between 7 and 10 wk of age, and the volume of the right testis remained small until 50 wk. In the adult male, the mean (±SEM) volume of the right testis was 870 ± 29.68 mm3. No sex difference in the age of onset of puberty could be discerned from our observations. In addition, our measurements did not indicate that there was a discernible spurt in growth around puberty in the marmosets.

Table 1 Mean (± SEM) physical measurements of adult marmosets
Full size table

Indices of care and growth. It is not clear whether caregiver-infant interactions are significantly influenced by the infant's gender. Thus, for correlational analysis we limited our investigation to females for which we had complete behavioral records and physical measures(n = 10).

The amount of parental care that a young marmoset received as an infant was significantly correlated with its body weight and physical stature as a juvenile. Specifically, the frequency that an infant was carried during its first 3 wk of life was correlated with its body weight at 10 and 20 wk (ρ= 0.75, p < 0.025 and ρ = 0.89, p < 0.007; Fig. 1A). Differential effects of early carrying were no longer apparent at 50 wk. The body weight of juveniles at 20 wk of age also was positively correlated with the frequency with which they were retrieved, groomed and explored socially by their caregivers during wk 1-3 (ρ = 0.85,p < 0.011; ρ = 0.69, p < 0.040; ρ = 0.66,p < 0.049, respectively; Fig. 1B). In addition, total care (care coefficient) was significantly correlated with juvenile body weight at 20 wk ( Fig. 1C; ρ = 0.72,p < 0.03).

Figure 1
figure 1

Body weight of juvenile females (20 wk old) is correlated with the frequency with which they were carried during infancy (wk 1-3) in panel A and with the frequency with which they were socially explored during infancy (wk 1-3) in panel B. (Spearman rank correlation: ρ = 0.544, p = 0.007 and ρ = 0.655, p= 0.05, respectively.) Frequency of total care during infancy (wk 1-3) is correlated with the body weight of juvenile females (20 wk old) inpanel C. Total care is assessed by subtracting the frequency of all of the punitive behaviors from the frequency of all of the positive behaviors(care coefficient). Spearman rank correlation: ρ = 0.724, p = 0.03. Some points represent the value for more than one animal.

Full size image

Not all caring behaviors were statistically correlated to all measures of physical stature. However, positive trends were observed in many measures by 10-20 wk. Correlations between early parenting received by the young marmoset and body weight and size at 3 wk was not significant. Similarly, measures of physical stature at 50 wk of age were no longer correlated with care received during wk 1-3. With the exception of direct parental abuse, the frequency of punitive behaviors was not significantly correlated with growth.

Abuse and growth and development. Juveniles and adolescents, regardless of sex, that experienced abusive caregiverinfant interactions during their first 3 wk of life were smaller in size than other juveniles. Figure 2 shows the growth curves for marmosets that were abused by their parents during infancy (n = 8) compared with those that were not abused (“had normal care”; n = 10). (Eight infants of the 26 did not have complete measures and were excluded from the analyses.) Marmosets that experienced negative parenting were smaller in body weight at 20 and 35 wk, knee-heel length at 50 wk, and head-tail length at 50 wk (p < 0.05; ANOVA followed by Fisher PLSD).

Figure 2
figure 2

Physical development in abused and nonabused (normal) animals as measured by body weight (g), knee-heel length (mm), head circumference (mm), and head-tail length (mm). Measures were taken at 3, 6,10, 20, 35, and 50 wk. (* = p < 0.05 vs normal infants; ANOVA followed by Fisher PLSD).

Full size image

Infants that experienced negative parenting also demonstrated different patterns of social development. Although between 4-6 and 7-10 wk infants that experienced normal care were more likely to be observed in proximity with their caregivers than those which had been abused (Mann-Whitney;U′ = 15, p < 0.03; Fig. 3) Between 11 and 20 wk, abused monkeys showed a greater frequency of proximity to their caregivers than normally reared infants (Mann-Whitney;U′ = 23.5, p < 0.02). By 50 wk of age, there were no differences in the frequency that normal or abused monkeys spent with their caregivers.

Figure 3
figure 3

Frequency with which abused and normal care infants were within proximity of their caregivers. Frequency is expressed as number of occurrences per 5-min observation time. Although normal care infants were proximal to their caregivers more frequently during wk 7-10, abused infants had a higher frequency of proximity between wk 11 and 20.

Full size image

Although huddling tended to be higher in abused infants between 4-6 and 7-10 wk of age, by 20 wk of age and through 50 wk, the frequency of huddling between abused animals and caregivers tended to be below normal (Fig. 4). Infants that experienced normal caregiver interactions during infancy played significantly more than abused infants between wk 7 and 10 (Mann-Whitney; U′ = 15, p < 0.03; Fig. 5), but at 20 wk of age, the differences in the frequency that normal or abused monkeys played with their caregivers were no longer statistically significant.

Figure 4
figure 4

Frequency with which abused and normal infants huddled with their caregivers. Frequency is expressed as number of occurrences per 5-min observation time. Abused infants tended to have a higher frequency of huddling between wk 4-6 and 7-10.

Full size image
Figure 5
figure 5

Frequency with which abused and normal infants played with their caregivers. Frequency is expressed as number of occurrences per 5-min observation period. Normal infants played with their caregivers more frequently during wk 7-10 than did abused infants.

Full size image

Abuse and pituitary-adrenal function. Mean basal plasma IR ACTH and IR cortisol levels were not different in young adults of an abusive parent compared with offspring of nonabusive parents. Similarly, no difference was seen in the incremental response of ACTH between the two groups. Cortisol responses to oCRH, however, were significantly lower in the young adults that had experienced negative parenting during infancy than those who had nonabusive parents (p < 0.05, unpaired t test;Table 2).

Table 2 Plasma IR ACTH and IR cortisol: basal values and responses to oCRH
Full size table

DISCUSSION

The results of the present study indicate that parental care received during infancy is correlated to subsequent growth and behavioral development in the common marmoset. The frequency of positive caring behaviors, many of which involved significant body contact between the caregiver and the infant, such as carrying, grooming, and social exploration, was significantly correlated with stature when the young monkeys reached 10 and 20 wk of age. Furthermore, young marmosets that were maltreated during infancy were significantly smaller in body weight and stature. In addition, those monkeys which had experienced negative rearing demonstrated abnormal patterns of social development. Finally, basal plasma ACTH and cortisol levels in the animals which had been abused were similar to those which had received normal care. However, although the ACTH response to exogenous oCRH was normal in the abused animals, the cortisol response was significantly blunted.

The absence of a care-stature relationship during early infancy suggests that the quality of “early care” affects “later” growth. On the other hand, the correlation between care and growth appeared to weaken as the animals approached puberty and adulthood (50 wk), suggesting that poor care from caregivers during early life led to a growth“delay” that was apparently “reversible,” with ensuing catch-up growth. These findings appear compatible with studies in humans where deficient parenting, including child abuse or neglect, also has been associated with deficient growth(28). Psychosocial dwarfism in humans is characterized by a normal birth weight, growth failure during childhood, and subsequent normal growth velocity with the potential for attaining normal adult height(29).

Our findings are compatible with studies in laboratory animals and humans who have shown that disruption of the child-caregiver relationship, as in abuse, contributes to marked physiologic and behavioral abnormalities in the offspring(30–36). Thus, in the present study the infants which had experienced negative parenting demonstrated overall impaired skeletal growth. The abnormal patterns of social development exhibited by these monkeys suggests an impairment of the ease with which social bonds are formed later in development. For example, between 11 and 20 wk, the frequency that abused marmoset infants were in the proximity of their caregivers increased to levels greater than that of normal infants. This behavioral pattern may suggest that abused animals are less securely attached to their caregivers, and is compatible with the findings of Chalmers and Locke-Haydon(23) who found that marmoset infants with a history of separation or unresponsive (tranquilized) caregivers, initially spent greater portions of time with each other than did controls, and subsequently, tried to climb onto their mother's back more frequently than did normal subjects. In humans, maltreated children have been characterized by insecure attachment and a tendency to exhibit anxious/avoidant attachment patterns(32, 33). Some of these “insecure attachment” children, however, have with time shown behaviors associated with more normal, secure attachments(32). Furthermore, when removed from their abusive environment, children display a compensatory increase in social behaviors that gradually return to normal levels(12). Girls sexually abused by their caregivers subsequently developed dysthymia or frank depression, as well as an abnormal ACTH response to oCRH stimulation(31). Similar phenomena may be seen in social separation studies in monkeys; upon reunion there is often a temporary increase in attachment behaviors(15).

The mechanism(s) which mediate the linkage between early care or emotional state and the physiologic and behavioral manifestations during development are not clear. The data presented in this study indicate that the parenting behaviors which correlated positively with growth, such as carrying, grooming, and social exploration, involved significant physical contact between the caregiver and the developing monkey. These contact behaviors may play a role in establishing a caregiver-infant bond, as well as in stimulating normal somatic growth. In this regard, it has been suggested that caregiver-infant bonding is a process which may depend upon species-characteristic behaviors which promote and/or maintain contact or proximity(34). On the other hand, physical contact and touch between young animals and their caregivers appears to be necessary for normal somatic growth(30, 31). In humans, the etiology of psychosocial growth retardation has been related to unsatisfactory mother-infant relationships, which is often measured in terms of physical contact(12, 35). In nonhuman primates, if physical contact is restrained while visual, auditory, and olfactory cues are maintained, the behavioral development in the animal remains abnormal(36).

It is now well known that the growth axis is inhibited at many levels during stress. In this regard, prolonged activation of the HPA axis leads to suppression of growth hormone secretion and inhibition of somatomedin C and other growth factor effects on their target tissues(37, 38). Thus, CRH stimulates somatostatin secretion, with resultant inhibition of growth hormone secretion, and chronically elevated glucocorticoids themselves inhibit GH secretion and cause resistance of target tissues to the actions of somatomedin C and other growth factors. However, our data suggest that the blunted cortisol response to exogenous oCRH occurs in the context of a hypofunctioning HPA axis in the monkeys that had experienced neonatal abuse. Although we do not understand the mechanism of this hypofunction, we speculate that it is associated with a generalized decrease in the activity of the stress system, which includes the noradrenergic system in the CNS(39–42). Noradrenergic input is crucial for GH secretion, and decreased noradrenergic activity would be expected to lead to decreased GH secretion and impaired somatic development. The dysregulation of the HPA axis expressed as hypofunctioning has been demonstrated previously in several human disorders, which include the posttraumatic stress disorder(43), the depressive phase of seasonal affective disorder(44), the period ensuing cessation of nicotine uptake(45), and hypothyroid patients and experimental animals(46, 47).

Our data indicate the impact of care, as measured by discrete, quantifiable behaviors directed by the caregiver toward the infant, on the development of the young. On the other hand, however, we cannot definitively exclude other factors which may have an impact on parental behavior, including the behavior of the infants themselves. Although our data do not document the potential multiple signals which may alter the quality of caregiving provided by the parents, it is known that, in marmosets, experience in rearing siblings is highly beneficial, but not essential, for the development of maternal behavior and reproductive success(48). Furthermore, we have observed that both inexperienced and/or stressed parents tended to kill or abuse their young(49). The parental inadequacy of these naive animals, however, seems to correct itself somewhat, as successive sets of infants are born(50).

In the study of this nonmanipulated system, we noted the spontaneous occurrences of abusive/neglectful behaviors, as well as caring/nurturant behaviors, and assessed whether these behaviors correlated with long-term growth and development. We observed a continuum in the quality of caregiver-infant interactions, and the effects of these on growth and development were, perhaps, not as robust as one might have expected in a manipulated system. Nonetheless, our data clearly suggest that parental support may significantly influence long-term physical and behavioral development, and that the absence of support or presence of abuse are associated with growth retardation and abnormal behaviors. Hence, it appears that the marmoset could provide a useful model both for understanding psychosocial short stature and for studying the mechanisms involved in the ensuing “catch up” growth, which may occur after correction of the negative environment. The hypoactivity of the HPA axis observed in the abused animals was compatible with the hypothesis that changes in the activity of the stress system may play a significant role in negatively influencing growth and behavior, as suggested above(39). Clearly, more hormonal and nutritional determinations are needed to clarify how the delayed growth observed was related to stress-induced hormonal changes and/or nutritional factors.