вход по аккаунту


Ecological constraints on the development of infant independence in rhesus.

код для вставкиСкачать
American Journal of Primatology 13:103-118 (1987)
Ecological Constraints on the Development of infant
Independence in Rhesus
Department of Enuironmental, Organismic and Population Biology, University of Colorado,
Mother-idant dyads were observed among three populations of rhesus monkeys (Macaca mulatta) in India and Nepal. This research had three purposes: l) to examine the influences of maternal activities (ie, feeding,
locomotion, and rest) and arboreality on the early development of infant
independence; 2) to determine whether interpopulation differences in the
development of infant independence can be accounted for by correlated
differences in maternal time budgets and the amount of time infants spent
on the ground; and 3) to explore how patterns of maternal care may or may
not mitigate increased risk of infant mortality resulting from human harassment. Despite significant interpopulation differences in maternal time
budgets, the amount of time infants spent on the ground, and the degree of
harassment the animals experienced, few interpopulation differences in
infant development could be found. The application of a multiple regression/
correlation analysis revealed that the development of infant independence
was affected by the above factors but that those factors were acting in
complementary ways to constrain development and t o minimize the differences across populations.
Key words: infant independence, mother-infant contact, maternal care, development,
maternal feeding, arboreality, rhesus, Mucucu mulattu
The present paper provides an analysis of the development of infant independence among free-ranging rhesus monkeys (Macaca mulatta) in India and Nepal.
This research had three purposes, namely: 1) to examine the influences, if any, of
maternal time budgets and the physical structure of the infant’s rearing environment on the early development of infant independence; 2) t o determine whether
interpopulation differences in the development of infant independence can be accounted for by correlated differences in adult time budgets andor the rearing
Received December 9, 1986; revision accepted April 14, 1987.
Address reprint requests to Dr. C.H. Southwick, Department of Environmental, Population and Organismic Biology, University of Colorado, Boulder, CO 80309.
0 1987 Alan R. Liss, Inc.
104 I Johnson and Southwick
environment; and 3) to explore how patterns of maternal care may or may not adapt
to mitigate an increased risk of infant mortality resulting from a disruption of the
animals’ habitat.
The possibility that interpopulation differences in the early development of
infant independence could be due to correlated differences in maternal time budgets
is suggested by the fact that, while foraging, rhesus and bonnet macaque (M.radiata)
mothers reject their infants’ attempts to nurse [Plimpton, 1981; Johnson, 19861.
Consequently, rhesus and bonnet infants are increasingly more likely to be separated from their mothers as their mothers’ foraging time increases [Plimpton, 1981;
Rosenblum and Sunderland, 1982; Johnson, 19861. Perhaps by discouraging her
infant’s nursing attempts during feeding episodes, a macaque mother is attempting
to maximize her foraging efficiency [Johnson, 1986; see also Altmann, 1980, 1983
for discussions of the same phenomenon in Papw cynocephalus]. If so, the more time
she must devote to feeding, the greater may be her need to encourage a degree of
autonomy in her infant. However, extensive group progressions requiring extended
periods of maternal locomotion could promote mother-infant contact since a dependent infant obtains transportation through contact.
Potentially, structural differences in the infants’ rearing environment could also
be responsible for interpopulation differences in development, particularly in rhesus
monkeys. Rhesus monkeys occur in a wide range of habitats, and the vegetational
structure of the habitat of any given group of animals is likely to determine in part
whether that group will be primarily arboreal or terrestrial. Since arboreality would
probably increase a primate infant’s risk of accidental injury or death due to
falls [Chalmers, 19721, terrestriality may facilitate the development of infant
Most previous studies examining the influence of environmental factors on
primate development, particularly those examining the influence of the physical
structure of the infant’s rearing environment, have been conducted within laboratory settings [eg, Rosenblum, 1974a,b]. This is understandable since laboratory
conditions permit the experimental manipulation of environments and infant experience. However, the lack of control afforded by field conditions can be compensated
for to some extent through choice of appropriate statistical analysis.
The data presented here were collected over four consecutive field seasons from
three populations of rhesus and were analyzed using multiple regressiodcorrelation
(MRC). MRC was particularly well suited to the analysis of our data since it allows
one to simultaneously examine the effects of many independent or predictor variables upon a single dependent variable [Cohen and Cohen, 19831. Thus, the degree
to which infant independence was influenced by variation in both the amount of
time mothers spent feeding and the amount of time infants spent on the ground
could be examined within a single regression analysis. MRC also made it possible to
obtain significance tests of the differences among the three study populations,
including a priori contrasts between specific populations, while simultaneously
bringing the effects of the above factors under “statistical control” [Cohen and
Cohen, 19831.
Two of the three rhesus populations studied are located in Nepal and the other
is in India. Since 1978, the Indian study population has experienced increased
human harassment and has exhibited an elevated rate of infant mortality [Johnson
and Southwick, 1984; Southwick et al, 19861. Anticipating that Indian rhesus mothers would, over time, adjust their patterns of interaction with their infants as a
result of this harassment, the mothers and infants of this population were sampled
twice, once during the birth season of 1981 and again 3 years later in 1984. The two
Kathmandu populations, subject to less harassment and exhibiting lower rates of
early infant mortality (see Study Populations below), were sampled once each in
Developmental Constraints in Rhesus / 105
1982 and 1983. It was anticipated that, as a consequence of increased maternal
solicitude on the part of the Indian rhesus, the infants sampled in India in 1984
would be comparatively less independent and thus would spend significantly more
time in contact with their mothers than would the infants sampled during the three
previous field seasons. Consequently, a priori contrasts were planned not only
between the 1984 and 1981data but also between the 1984 data and those obtained
during the 1981,1982,and 1983 field seasons taken collectively.MRC enabled us to
make these contrasts free of the potentially confoundingeffects that could have been
introduced by interpopulation differences in the amount of time mothers spent
feeding and/or the degree of terrestriality their infants experienced.
Study Sites
The study sites have been described in detail elsewhere [Johnson and Southwick, 19841. The three locations represent characteristic habitats of rhesus macaques in India and Nepal: 1)Chhatari is a typical roadside and rural agricultural
setting located 22 km north of Aligarh, Uttar F’radesh; 2) Swayambhu is a temple
and parkland setting located to the west of Kathmandu, Nepal, approximately 700
km east of Chhatari; and 3) Gaushalla is a reserve forest patch, also located in
Kathmandu valley, 5 km east of Swayambhu. As indicated above, the monkeys at
Chhatari were sampled twice, once during the summer of 1981 and once during the
summer of 1984. The rhesus monkey populations at Swayambhu and Gaushalla
were sampled during the summers of 1982 and 1983, respectively.
Study Populations
The mother-infant sample was drawn from two rhesus groups at both Chhatari
and Gaushalla and from six groups at Swayambhu. There was a total of 145
monkeys at Chhatari in 1981, and despite a high rate of mortality among infants at
this site (see below), the number had increased to 191 by the summer of 1984. The
rhesus population at Swayambhu in 1982 consisted of 322 animals, 212 of whom
were members of the six study groups. Finally, one of the two rhesus groups initially
occupying the Gaushalla forest in 1983 left the area early in the 1983 field season.
TABLE I. Weekly Totals of Accumulated Observation Time (minutes) and the Number of Infants
Observed at Each of the Three Study Sites
Age (wks):
No. Infants
No. Infants
No. Infants
No. Infants
106 I Johnson and Southwick
Consequently, a complete census was never made of this group. It consisted of at
least 90 animals, however. Thus it was slightly larger than the remaining Gaushalla
group, which was composed of 81 monkeys.
In September of 1978 the larger of the two groups at Chhatari (Chhatari A) was
displaced from its place of residence, the area surrounding a small rural school yard,
by school officials and was subsequently forced into three parallel lines of trees
bordering the Aligarh-Anupshar road. Previously, these trees had been occupied
only by the smaller of the two rhesus groups at Chhatari (Chhatari B). As a result
of the displacement, the home range of both groups was diminished, and this in turn
made it more difficult for all of the animals to avoid harassment from the local
Since 1959, census data have been collected on all Aligarh district rhesus
groups, including the two groups at Chhatari [Southwick et al, 19801. Prior to the
displacement of Chhatari A, early infant mortality (ie, mortality occurring within
the first 4 months of infant life) for both groups averaged just 5.4%. Early infant
mortality increased fourfold to 22.7% during the 1979 birth season, the first following the displacement. Early mortality at Chhatari for 1980 through 1984 was 21.4,
11.1, 15.4, 1.8, and 18.0%. Since 1978, infant mortality for all other Aligarh district
groups, which were not subject to a home range displacement but have experienced
the same weather conditions as the Chhatari rhesus, has been consistently lower,
averaging 13.6,0.0,3.6,9.1,0.0, and 12.0%from 1979to 1984,respectively. However,
during these years the absolute percentage difference in infant mortality between
the Chhatari rhesus and the other Aligarh groups declined from a high of 21.4% in
1980 to a low of just 1.8%in 1983.
Early infant mortality for the Swayambhu and Gaushalla rhesus populations
has also been lower than that at Chhatari since the displacement, ranging from 3 to
8% from 1974 to 1977 [Teas, 19781. In 1982 early infant mortality was again only
about 2% for the Swayambhu monkeys. Furthermore, no infant deaths occurred
within any of the study groups at either Swayambhu or Gaushalla during the 1982
and 1983 field seasons.
The method of data collection has been described by Johnson and Southwick
[1984]. Briefly, four thirty-minute focal samples [Altmann, 19741 per week were
sought for 9 to 12 rhesus mothers (each presumed to be multiparous) and their
infants. Two afternoon and two morning observations were sought between 0500
and 1800 hr,with no more than one observation being conducted per day. However,
the 2-hour-per-weekgoal for data collection for each dyad was not met in all cases.
A total of 523 hr of direct observation was accumulated over the four field seasons.
The number of infants and the total accumulated observation time per week of
infant life are presented in Table I. Most infants were of known age; however,
included within each infant sample were a number of infants born before the arrival
of the first author at each site. Although these infants were 3 weeks of age or
younger when their ages were estimated, they ranged from 1to 5 weeks of age by
the time data collection began. Their inclusion within the infant samples meant
that during any given week, infants of several different ages were being observed,
and the effects of weather (eg, premonsoon temperatures in excess of 40°C at
Chhatari) upon any particular age group were thus minimized.
At the onset of each observation, the ongoing behaviors of the focal dyad were
recorded, and, using a digital watch, the time at which all subsequent behaviors
began and ended were noted in code. Thus a cumulative record of the total time both
mother and infant spent engaged in each behavior was generated. However, when a
mother and her infant were separated, obtaining an accurate record of the infant’s
Developmental Constraints in Rhesus / 107
TABLE 11. Definitions of Behaviors
Ventro-ventral contact
Nipple contact
Total contact
Maternal feeding
Maternal locomotion
Maternal rest
Maternal protection
Maternal rejection
Ground time
Ventral surfaces of mother and infant juxtaposed
Infant orally contacting either of its mother’s nipples (may
occur independently of, or concurrently with ventroventral contact)
Physical contact of any form between mother and infant
(subsumes all of ventro-ventral and nipple contact)
Collection and ingestion of food (mere chewing was not
Mother walking or climbing (if occurring simultaneously
with Maternal Feeding, only Feeding was recorded)
Any maternal activity other than feeding or locomotion
Mother maintaining or regaining contact by either
restraining the infant from breaking contact, or
retrieving the infant and contacting it to her ventrum
Mother biting or hitting her infant, forcibly terminating
nipple contact, preventing the infant from contacting her
nipple, or breaking contact within 5 seconds after the
infant initiates contact with her
Infant on the ground or in contact with its mother who is
on the ground (may occur independently of, or
concurrently with, any of the three categories of contact)
Change in the behavior of either mother or infant in
response to approaching domestic dogs or to humans
throwing stones at the animals (two potential sources of
infant mortality)
behavior was often not possible, particularly when the infant was located in dense
vegetation. Owing to the mother’s larger size and greater observability, her behavior
was given a greater priority than that of her infant’s.
The birth season peaks at all three sites in May and June. Consequently,
observations were conducted during the months of June, July, and August, and the
development of mother-infant relations was observed during the first 11 weeks of
infant life. The monkeys at each site are in daily contact with people; hence, the
animals were already well habituated to close observation.
The analyzed behaviors are defined in Table It.All measures other than maternal protection, maternal rejection, and disturbance were expressed as a percentage
of the total time a mother-infant pair was observed during an observational week.
In contrast, the frequency of disturbance incidents at each site was summed over
the entire field season, while instances of maternal protection and maternal rejection recorded during any given week were summed and expressed as a rate.
It has long been recognized that any given index of development is not likely to
be as applicable to some species as it is to others [Mack, 19791. However, even within
species it is unlikely that any two indices will be of equal significance to mothers
and infants [Nash, 19781. The three categories of mother-infant contact defined in
Table I1 have all been used previously by other investigators to index the development of independence in cercopithecine infants [Rosenblum, 1974a; Altmann, 1980;
Berman, 1980; Lee, 19841, but this is the first study to simultaneously consider all
three and to examine to what extent they may be affected differently by environmental variables.
108 I Johnson and Southwick
TABLE 111. Comparisons of Time (%) Mothers Spent Feeding or
Locomoting, and Time Infants Spent on the Grounda
Maternal feeding
Infant sample
Gaushalla Swayambhu
Maternal locomotion
Ground time
aUnderlined means indicate no significant difference. Contrasts made using the Scheffe
Statistical Analyses
The arcsine transformation was employed to equalize the variances of all percentage measures (eg, total contact), with both dependent and independent variables
alike being transformed prior to all MRC analyses. All regressiodcorrelation statistics reported here are based on the transformed data. However, in order to facilitate
comparison with other studies, reported means, both in tables and in figures, are
based on untransformed data.
For the three measures of mother-infant contact (ie, ventro-ventral, nipple, and
total contact), the primary strategy of analysis was to apply MRC simultaneously to
the data from all four field seasons. The unit of statistical analysis was the infantweek; hence, the data from a given mother-infant pair were represented in the MRC
analyses as n data points, with n equal to the number of weeks that particular
mother-infant pair was observed during a field season. Because the number of weeks
individual infants were observed varied from a low of two to a high of ten, no
attempt was made to assess the magnitude of between-subject and within-subject
variance. Rather, each infant-week was treated as an independently derived data
point. The appropriateness of this procedure is addressed in the Discussion.
In all MRC analyses, the four infant samples were handled as a single nominal
research factor. However, a nominal factor consisting of four mutually exclusive
categories can be understood as having just three “aspects” and, as such, must be
represented in an MRC analysis as a set of three independent variables for purely
structural reasons (ie, in accordance with the form of the nominal research factor)
(Cohen and Cohen, 19831. There are several different methods by which the nominal
factor “infant sample” could have been represented. “Effects” and “dummy” coding
were chosen since these encoding schemes enabled us to obtain a priori significance
tests of the difference between specific infant samples or groups of samples while
simultaneously controlling for the effects of other predictor variables [see Cohen and
Cohen, 1983 for a discussion of these procedures]. The three independent variables
encoding the four infant samples were entered into the MRC analyses en masse,
Developmental Constraints in Rhesus / 109
l O O r
l O O r
6 - nC H H A T A R I
(3-4G A U S H A L L A
&-A C H H A T A R I 1'94)
Fig. 1. Mean percentages of time spent by mother-infant pairs in ventro-ventral, nipple, and total contact.
110 I Johnson and Southwick
C H H A T A R I (‘81)
I * I
1 x 1
Fig. 2. Mean rates of maternal protection. *, Mann-Whitney U test comparing the data from the 1981 and
1984 field seasons a t Chhatari, P < 0.05. (a)Mann-Whitney U test comparing the data from the 1984 field
season with all others taken collectively, P < 0.05.
while all other independent variables were entered in a stepwise fashion on the
basis of their partial correlations with the dependent variable.
The application of MRC simultaneously to data representing infants of a variety
of ages was accomplished by entering the age of the infants as an independent
variable in the analysis. This strategy of making interpopulation comparisons represents a methodological departure from previous studies of primate mother-infant
relations [eg, Chalmers, 1972; Rosenblum, 1974a,b; Altmann, 1980; Lee, 19841,
which have applied statistical tests independently to the data corresponding to each
week (or month) of infant life in an effort to detect age-specific differences in
behavior. However, a week-by-week analysis of the data for the three measures of
contact was also conducted by means of a one-way analysis of variance, an analysis
comparable in rationale to that conducted by previous investigators.
Analysis of variance was also used to test for differences among populations in
maternal feeding and ground time, while pair-wise comparisons were made employing the Scheffb procedure with the “experiment-wise” alpha set at 0.05. Variances
for maternal protection and maternal rejection proved strongly heterogeneous; hence,
these behaviors were analyzed employing the Mann-Whitney U test. Finally, the x2
test was employed to test for significant differences in disturbance among the study
Interpopulation Comparisons
The mean percentage of time infants spent on the ground ranged from a high of
82.4% at Swayambhu to a low of 39.8% at Chhatari. In contrast, maternal feeding
ranged from just 3.5% to a high of only 6.5%, while maternal locomotion ranged
from 8.3 t o 12.3%(Table 111).The extremely low values for maternal feeding are the
result of the restrictive definition of feeding employed; the animals spent considerably more time manipulating potential food items, as well as processing and reingesting foods removed from their cheek pouches. However, despite the differences in
magnitude, all three variables varied significantly over the four field seasons (ground
time, F[3,266]=72.77, P < 0.001; maternal feeding, F=5.67, P< 0.001; maternal locomotion, F = 11.32, P < 0.001). Pair-wise comparisons indicated that the Swayambhu
Developmental Constraints in Rhesus / 111
infants spent significantly more time on the ground than did the infants of the three
other samples, and that the rhesus mothers of Gaushalla were the most deviant in
terms of the time they spent feeding. Interpopulation variation in maternal locomotion was fairly uniform, however (Table Inn.
Disturbance also varied significantly with habitat (Table IV).Unlike the data
on maternal feeding and ground time, the data on disturbance at Chhatari are
derived solely from the 1984 field season; unfortunately, disturbance incidents were
not classified as to their source during the summer of 1981. Expected frequencies
were calculated on the assumption that the frequency of such incidents should have
been proportional to the amount of time each population was observed.
The rate of disturbance at Chhatari was over two and one half times greater
than that at Swayambhu and over three times greater than that at Gaushalla (Table
IV). The human residents of the Chhatari area view the monkeys as agricultural
pests and, in fact, devote considerable energy to the task of keeping the monkeys
out of the cultivated fields that surround the trees now occupied by both Chhatari
rhesus groups. During the course of this study, humans were observed to chase and
throw stones at the monkeys at all three study sites but, owing to Chhatari’s
agrarian circumstances, such incidents were a far more integral part of the daily
routine at this site than at the others for monkeys and humans alike.
Despite the above interpopulation differences in maternal feeding, ground time,
and disturbance, the development of mother-infant contact varied little across populations Pig. 1).Indeed, none of the weekly differences among the infant samples
for the three contact measures proved significant when tested by means of a oneway analysis of variance. Although no significant age-specific differences in the
development of mother-infant contact were detectable, small but consistent differences among the infant samples were apparent. For example, during seven of the
infants’ first eleven weeks, the 1984-Chhatariinfants spent more time contacting
their mothers’ nipples than did the infants of any other sample. Their TOTAL
CONTACT scores were also the highest or second highest for all but two of the
eleven weeks.
In order to determine whether the weekly differencesamong the infant samples
would attain statistical significancewhen considered collectively,the effect of infant
age was partialled through linear regression, and the residuals of the four infant
samples were examined for significant heterogeneity. Once the effect of infant age
had been partialled, significant differences were demonstrable in the observed
amounts of nipple contact (Table V), but the observed differences in total contact
and ventro-ventral contact still failed to attain statistical significance. In order to
isolate the consequences of arboreality from the effects of differing maternal time
budgets, each form of contact was then simultaneously regressed on maternal
feeding, maternal locomotion, ground time, and infant age.
Both maternal feeding and ground time were found to be significantly and
negatively correlated with all three measures of mother-infant contact (Table VD.
TABLE IV. Observed and Expected Frequencies of Disturbancea
Chhatari (‘84)
Observed frequency
Expected frequency
Rate (freqh)
X2 = 77.0 P < 0.005
aExpectedfrequencies calculated assuming that such incidents occurred proportionately to the amount of
time infants were observed at each site.
112 I Johnson and Southwick
TABLE V. Apparent Heterogeneityof the Infant Samples for Each Contact Measure (as
Evidenced by the Corresponding F-ratio)Before and After the Effects of Correlated
Variables Were Held Constant
Contact measure
Ventro-ventral contact
Nipple contact
Total contact
With only
infant age
(df = 3,265)
With maternal
feeding and ground
time partialled
(df = 3,263)
With maternal
rest and ground
time partialled
(df = 3,263)
*P < 0.05.
**P < 0.01.
Their partial correlation coefficients indicate that the probability that mothers and
their infants would be in contact decreased as either maternal feeding time or infant
ground time increased. In contrast, maternal locomotion was not found to correlate
with any form of mother-infant contact. Of greater importance, however, was the
fact that, once the effects of ground time and maternal feeding were held constant,
significant heterogeneity was demonstrable for total contact but not for nipple
contact, a reversal of the previous status of these two variables (Table V).
When the data for each field season were examined separately, mother-infant
contact was still found to be inversely correlated with both maternal feeding and
ground time (although the partial correlation coefficients did not attain statistical
significance). Thus, it may be concluded that the associations were not the result of
repeated observations of idiosyncratic mother-infant pairs within one or two infant
samples. Rather, it appears that the effects of ground time and maternal feeding
were operative across field seasons.
The unequal numbers of male and female infants in the four infant samples
might have contributed to the variance of the data. Consequently, the regression
analyses discussed above were repeated with the sex of the focal infants also entered
as an independent variable. Although male infants spent, on the average, less time
contacting their mothers than did female infants, the differences between male and
female infants were in no case significant. Thus the differences in total contact
demonstrable after the effects of maternal feeding and ground time were held
constant could not be attributed to differences in the relative number of female and
male infants constituting the four infant samples.
The fact that significant heterogeneity in total contact was evident only after
the effects of ground time and maternal feeding were held constant indicates that
variation in one or both of these variables was suppressing the expression of other
differences among the infant samples, ie, differences due to interpopulation variation in at least one other factor not included in the analysis. The heterogeneity of
the total contact scores can be attributed to interpopulation variation in disturbance
(see "A Priori Contrasts" below) and, inasmuch as holding the effect of variation in
maternal feeding constant acted t o increase the apparent heterogeneity of the
ventro-ventral, nipple, and total contact scores while partialling the effect of ground
time decreased heterogeneity, we may conclude that it was maternal feeding that
was acting to suppress the effect of disturbance. Moreover, interpopulationvariation
in maternal feeding was also suppressing the effect of ground time, making the
infant samples appear more similar than they would have if the infants' mothers
Developmental Constraints in Rhesus / 113
had devoted equal amounts of time to feeding or had been exposed to equal amounts
of disturbance. Finally, since no significant heterogeneity in nipple contact was
detectable after the effect of ground time was partialled, we may also conclude that
the observed differences in nipple contact were attributable to correlated differences
in the amount of time infants spent on the ground, but, to reiterate, this was not
true for contact as a whole.
For infants t o exhibit an increased probability of being out of contact with their
mothers during one maternal activity, such as feeding, they obviously must also
exhibit an increased probability of being in contact during at least one other maternal activity. Because contact did not correlate with maternal locomotion, infants
must have experienced greater amounts of contact when their mothers were resting
and, if mothers preferentially rested on structures above ground level, then the
inverse correlation between GROUND TIME and contact could have been spurious.
Maternal rest was, indeed, inversely related to ground time (df = 268, r =
-0.121, P = 0.06). To test whether the correlation between contact and ground time
was the result of the mothers’ predilection to rest when not on the ground, each
form of contact was regressed on infant age, ground time, and maternal rest [maternal rest was not included in the previous analysis due t o problems with multicolinearity, see Cohen and Cohen, 19831. As expected, the partial correlation between
maternal rest and mother-infant contact was both positive and significant, but the
partial correlation coefficient for ground time, though smaller than before, remained
significant (Table V). It can be concluded that the effect of decreasing ground time
on mother-infant contact was not merely the result of the mothers’ preference for
resting on certain substrata rather than others.
Substituting maternal rest for maternal feeding in the above regression analyses demonstrated one other salient point, namely, that significant heterogeneity
was demonstrable among the TOTAL CONTACT scores after the effect of either of
these variables had been partialled (Table V). Thus in every respect maternal rest
proved to be as good a predictor of mother-infant contact as did maternal feeding.
A Priori Contrasts
Because significant heterogeneity among the infant samples was demonstrable
for total contact but not for ventro-ventral or nipple contact after the effects of
correlated variables were held constant, it was appropriate t o use this measure to
obtain significance tests of 1)the differences between the data from the 1981 and
1984 field seasons at Chhatari and 2) the difference between the 1984 data and the
data for all other field seasons taken collectively.
TABLE VI. The Partial Correlation Coefficients Between Dependent and Independent
Variables and Their Statistical Significance. In Each Case the Effects of Infant Age Have
Also Been Partialled
Dependent variable
Nipple contact
Total contact
*P < 0.05.
**P < 0.01.
***P < 0.001.
Independent variable
Ground time with feeding
Maternal feeding (or rest)
with ground time partialled
(or rest) partialled
-0.25*** (0.24***)
-0.30*** (0.33***)
-0.29** * (0.26** *)
-0.26*** (-0.23***)
-0.27*** (-0.22***)
114 I Johnson and Southwick
The two a priori tests yielded differing results. As predicted, the infants sampled
in 1984 were found to spend significantly more time in contact with their mothers
than did the infants sampled at Chhatari in 1981once the effects of maternal feeding
and ground time were held constant (F = 7.85, P <0.005). However, there was no
significant difference demonstrable between the dyads sampled at Chhatari in 1984
and those of the other three infant samples considered collectively (F = 3.12,
P <0.08).
Maternal protection and rejection are the most obvious means by which rhesus
mothers regulate contact with their infants. Thus these measures provide a means
of gauging the relative solicitude of the mothers sampled over the four field seasons
and identifying the origin of the difference between the 1981 and 1984 total contact
Unfortunately, the expression of MATERNAL PROTECTION was so strongly
age specific (Fig. 2) that it was not appropriate to pool the data from all 11weeks as
in the previous analyses. Week-by-weekcomparisons were possible, and, when the
1984 Chhatari sample was compared directly to the 1981Chhatari data, significant
differences were demonstrable for weeks 2, 4, 8, and 9 (Mann-Whitney U test, P
<0.05). During each of these weeks, the 1984 Chhatari mothers restrained and
retrieved their infants more than the 1981Chhatari mothers did. Significant differences were also found for weeks 8 and 9 when the 1984 Chhatari sample was
compared to all others taken collectively (Mann-Whitney U test, P < 0.05).
No relationship was detectable between maternal protection and infant sex.
However, the major determinant of maternal rejection rates was, in fact, the sex of
the infant. Male infants received maternal rejections at rates exceeding those for
females during 10 of the first 11weeks (sign test, P <0.01). Because of the association between maternal rejection rate and infant sex, and because a majority of the
infants sampled at, Chhatari in both 1981 and 1984 were females, comparisons
between sites were made for female infants only. This diminished the power of all
subsequent significance tests but, since there was little association between maternal rejection rate and age for female infants [Johnson, 19851, it was possible to
combine the data from the full 11-weekperiod.
The pooled data illustrate just how infrequent maternal rejection was at all
three study sites. The 1981 Chhatari infants were rejected the least, receiving just
0.7 rejections per hour. In contrast, the Swayambhu, Gaushalla, and 1984 Chhatari
infants received maternal rejections at the rate of 1.4, 1.7, and 1.4 per hour, respectively. The difference between the two Chhatari samples proved significant (MannWhitney U test, P <0.01), but, not surprisingly, the difference between the 1984
Chhatari females and those of the other three samples taken together did not.
The results of this study indicate that, among free-ranging rhesus monkeys,
mother-infant contact is positively correlated with increasing arboreality but inversely associated with maternal feeding time. However, despite the fact that the
amount of time rhesus mothers and infants spent in contact was affected by environmental factors, the general course of development varied little across populations
(Fig. 1). Indeed, no significant differences could be found among the infant samples
when the developmental curves for nipple, ventro-ventral, and total contact were
examined on a week-by-weekbasis.
In an earlier paper [Johnson and Southwick, 19841, we presented data from the
first three field seasons of this study, data that had not yet been subjected to an
MRC analysis. Because mother-infant contact developed similarly across the Indian
and Nepalese study populations despite significant interpopulation differences in
Developmental Constraints in Rhesus / 115
infant ground time (Table III), we concluded that “environmental structure as
observed in this field study exerts little influence upon early infant development” (p
108). It is only with the addition of data from a fourth field season and, more
important, the application of an MRC analysis to the entire data set that we were
able to discern the developmental consequences of increasing arboreality in rhesus.
However, implicit in all of the MRC analyses presented above and discussed
below is the assumption that the infant-week is an appropriate unit of statistical
analysis and that treating repeated measurements from individual mother-infant
pairs as independent data points did not unduly affect estimated P values. We feel
that the use of the infant-week was appropriate, particularly in a first attempt to
assess the efficacy of MRC in evaluating the influence of multiple environmental
factors on the development of infant independence. Furthermore, we feel that the
partial correlations of both maternal feeding and ground time with mother-infant
contact were sufficiently strong (Table VD and the independence assumption sufficiently robust to conclude that the associations between these variables would
remain significant even if the relative magnitude of within- and between-subject
variance could be assessed. The fact that the effects of ground time and maternal
feeding acted to suppress one another emphasizes the need of a multivariate approach to the study of behavioral development in free-ranging primates. Unfortunately, mother-infant samples much larger than those typical of most field studies
[eg, Altmann, 1980; Berman, 1980; this study] would be required to employ multivariate procedures on an age-specificbasis.
The mutually inhibitory effects of ground time and maternal feeding suggest
that the development of infant independence may be constrained by environmental
factors rather than unaffected by them as we originally concluded. For example,
both the Indian and Nepalese rhesus mothers of this study actively discouraged
their infants’ attempts to maintain contact during feeding bouts [Johnson, 19861
and, although arboreality promoted mother-infant contact, the location of some foods
(eg, the fruit and leaves of trees) and the mothers’ intolerance of contact coincident
with feeding may have worked together to set an upper limit to the amount of
contact infants could obtain when not on the ground. As a consequence of the
complementary action of environmental factors such as those above, the degree to
which rhesus mothers may be able to adjust their patterns of maternal care to
mitigate an increased risk of infant mortality, as at Chhatari, may be limited.
As expected, rhesus infants at Chhatari did spend significantly more time in
total contact with their mothers during the summer of 1984 than they did in 1981,
and the greater protectiveness exhibited by Chhatari mothers in 1984 suggests that
the difference between years was due to changes in the mothers’ behavior. Furthermore, the fact that the difference between the level of infant mortality occurring
among the Chhatari rhesus and that occurring among all other Aligarh monkey
groups declined from a high of 21.4%in 1980 to just 1.8%in 1983 and 6.0%in 1984
suggests that the adjustment made by the Chhatari mothers may indeed have
diminished risk of injury to their infants. However, total contact among the 1984
Chhatari dyads was not significantly greater than that occurring within the other
three infant samples taken collectively, indicating that, despite the Chhatari mothers’ greater solicitude, the development of independence among the 1984 infants
was still within species’ norms. Indeed, the fact that no significant differences in
nipple or ventro-ventro contact were demonstrable once the potentially confounding
effects of ground time and maternal feeding were held constant emphasizes further
how small the developmental differences among the infant samples really were, and
supports the suggestion that there were limits imposed upon the Chhatari mothers’
ability to adjust their pattern of caregiving.
116 I Johnson and Southwick
The developmental curves for the 1984 Chhatari infants may not have been
notably deviant in part because their mothers not only restrained and retrieved
them more frequently than did the 1981 Chhatari mothers but they also rejected
them more. A resolution of this apparent contradiction may be found in Altmann’s
[1980] suggestion that maternal rejection serves as much to coordinate mother and
infant activities as it does to encourage infant independence in general. Perhaps
because of the increased protectiveness of the 1984 Chhatari mothers and the
increased levels of contact that it promoted, these mothers had to reject their infants
more in order to successfully coordinate their own maintenance activities with their
infants’ demands for contact. This supposition remains purely conjectural but it
emphasizes the conflicting pressures imposed upon rhesus mothers during their
infants’ period of dependency.
The negative association between all three forms of contact and maternal feeding was matched by an equally strong but positive correlation between motherinfant contact and maternal rest (Table VI); the two correlations represent complementary manifestations of the same phenomenon, namely, the probability that
rhesus mothers and infants will be in contact is dependent in part upon the mothers’
ongoing behavior. However, both maternal rest and maternal feeding would probably have accounted for a larger proportion of the variance in the contact scores if
they had been defined differently. Maternal rest included a diversity of maternal
activities (eg, grooming, sleep, and the manipulation of nonfood objects), and it
seems reasonable to suppose that rhesus mothers may be as rejective of contact
coincident with some of these activities as they are of contact coincident with
feeding. Consequently, maternal rest would probably have served better as a predictor of mother-infant contact if it had been defined more narrowly. Similarly, maternal feeding would probably have served better as a predictor of mother-infant contact
if it had included the mother’s search for food rather than just her collection of it,
since foraging in general is likely to be as adversely affected by infant contact as is
feeding in particular. Future research would benefit from a refining of the definitions of maternal activities, since improving their predictive power would decrease
the remaining variance of the data and enhance our ability to detect small interpopulation differences in infant development. We will return t o this point below.
The source of the inverse correlation between mother-infant contact and ground
time is not as easily identified as that for maternal feeding. The association is
consistent with Chalmers’ [1972] suggestion that arboreality should promote motherinfant contact since such contact would diminish the infant’s risk of accidental
death. If arboreality is assumed to be frightening to infants, the negative association
between mother-infant contact and ground time is also consistent with the suggestion of some laboratory investigators [eg, Jensen et al, 1968; Rosenblum, 1974a,b]
that environmental structure or “complexity” should intimidate young infants and
inhibit independent action. However, ground time was more strongly associated
with ventro-ventral contact and nipple contact than it was with total contact (Table
VD, suggesting that decreasing amounts of ground time promote specific modes of
contact rather than contact per se.
Rhesus monkey infants under three months of age only seldom consume solid
foods [Kaufmann, 1966; Johnson, 19851, and, since nutritionally dependent infants
must recoup their energetic losses through suckling, one would expect that arboreality would be most closely associated with nipple contact and ventro-ventral contact
if decreasing ground time promoted infant fatigue. While the branches of bushes
and most trees appeared negotiable by all but the very youngest of infants (ie,
infants less than two weeks old), all infants clearly had difficulty in locomoting over
substrata other than the ground, and no infant under 11weeks of age was capable
of climbing or descending a mature tree unaided. Thus arboreality did impede infant
locomotion and probably made it energetically more costly as well.
Developmental Constraints in Rhesus / 117
The fact that infants had to be carried into and out of larger trees undoubtedly
contributed to the association between arboreality and mother-infant contact. Indeed, the fact that a rhesus mother typically carries her infant on her ventrum with
the infant orally contacting one of her nipples would also contribute to ground time’s
stronger association with ventro-ventral and nipple contact than with contact as a
whole. However, the maternal activity associated with the time spent above ground
level was rest, not locomotion. Moreover, when each category of contact was regressed simultaneously on maternal locomotion and ground time, only the latter
variable proved able to account for a significant proportion of the variance in contact
scores. Thus we may conclude that the negative association between ground time
and contact was not merely the manifestation of the mothers’ willingness to carry
their infants when not on the ground [see also Johnson, 19863. Similarly, we may
also conclude that the differences among the partial correlations between ground
time and each of the three categories of contact did not result solely from the infants’
inability to climb trees.
One final point is appropriate here. Investigators should be mindful of the fact
that one’s conclusions concerning ecological constraints upon development may vary
as a function of the number of independent variables included in one’s analysis and/
or the manner in which infant independence is indexed. Clearly, we would have
come to quite different conclusions about the influence of human harassment on
mother-infant contact had we not simultaneously considered the effects of arboreality and maternal time budgets, or if we had utilized only one index of development.
It is also possible that if maternal rest or maternal feeding had been defined
differently, we might have been able to detect some effect of human harassment on
nipple and ventro-ventral contact, not just total contact. Because environmental
factors affect different indices of development differently, studies reporting a lack of
an association between development and an environmental factor should be interpreted with care.
1.Maternal time budgets and the structure of the rearing environment can and
do affect the early development of infant independence in rhesus macaques; however, in the present case, these factors acted in complementary ways to maintain
the stability of early development rather than to promote interpopulation differences.
2. Chhatari infants spent significantly more time in contact with their mothers
in 1984 than in 1981, a change in behavior probably related to the high levels of
disturbance to which the animals at this site were subjected. This is a specific
example of how primate mothers can adjust their pattern of caregiving so as to
mitigate an increased risk of infant mortality, but the scope of such an adjustment
appears likely to be limited by other environmental factors constraining the course
of infant development.
This study was supported by USPHS grant RR-01245 and NSF grant BNS830825 to the University of Colorado. David Armstrong, Marc Bekoff, Alexander
Cruz, and David Greene provided critical comments during the course of the study,
and M. Farooq Siddiqi provided logistical support in India. This study forms a part
of the doctoral dissertation of the first author.
FANTS. Cambridge, Harvard University
Press. 1980.
Altmann, J. Observational study of be- Altmann, J. Costs of reproduction in baboons
haviour: sampling methods. BEHAVIOUR (Pupio cynocephulus). Pp 123-145 in BEHAVIORAL ENERGETICS: THE COST OF
118 I Johnson and Southwick
pey; S.I. Lustick, eds. Columbus, Ohio, Ohio NORTHERN NEOTROPICS. J.F. Eisenberg, ed. Washington, DC, Smithsonian InState University Press, 1983.
Berman, C.M. Mother-infant relationships stitution Press, 1979.
among free-ranging rhesus monkeys on Nash, L.T. The development of the motherCay0 Santiago: a comparison with captive infant relationship in wild baboons (Papi0
pairs. ANIMAL BEHAVIOUR 28:860-873, anubis). ANIMAL BEHAVIOUR 26:7461980.
759, 1978.
Chalmers, N.R. Comparative aspects of early Plimpton, E. ENVIRONMENTAL VARIinfant development in some captive Cercop- ABLES AND THE RESPONSE TO MAithcines. Pp 63-82 in PRIMATE SOCIALI- TERNAL LOSS. PhD dissertation, Suny
ZATION. F. Poirier, ed. New York, Random Downstate Medical Center, New York, 1981.
Rosenblum, L.A. Differences in mother-inHouse, 1972.
Cohen, J.; Cohen, P. APPLIED MULTIPLE fant attachment in monkeys. Pp 123-145 in
SIS FOR THE BEHAVIORAL SCIENCES. Freedman; R.M. Riehart; R.L. Van de Weile,
eds. New York, John Wiley, 1974a.
Hillsdale, New Jersey, Erlbaum, 1983.
Jensen, G.D.; Bobbitt, R.A.; Gordon, B.N. Ef- Rosenblum, L.A. Sex differences, environfects of environment on the relationship be- mental complexity, and mother-infant relatween mother and infant pigtailed monkeys tions. ARCHIVES OF SEXUAL BEHAV(MUCUCU nernestrina). JOURNAL OF COMIOR 3:117-128,1974b.
PARATWE AND PHYSIOLOGICAL psy- Rosenblum, L.A.; Sunderland, G. Feeding
CHOLOGY 66:259-263,1968.
ecology and mother-infant relations. Pp 75Johnson, R.L. MOTHER-WFANT RELA- 110 in PARENTING ITS CAUSES AND
INDIA AND NEPAL. phD delman; H.R. Schiffman, eds. Hillsdale, New
Erlbaum, 19g2.
dissertation, University of Colorado, BodSouthwick, C.H.; Richie, T.; Taylor, H.; Teas,
der, 1985.
Johnson, R.L. Mother-infant contact and H.J.; Siddiqi, M.F. Rhesus monkey populamaternal maintenance activities among tions in India and Nepal: patterns of growth,
free-ranging rhesus monkeys. PRIMATES decline, and natural regulation. Pp 151-170
Johnson, R.L.; Southwick, C.H. Structural di- LATION REGULATION. M.H. Cohen; R.S.
versity and mother-infant relations among MalPass; H.G. Klein, eds. New Haven, Yale
rhesus monkeys in India and Nepal. FOLIA University Press, 1980.
PRIMATOLOGICA 43:198-215,1984.
Southwick, C.H.; Siddiqi, M.F.; Johnson, R.
Kaufmann, J.H. Behavior of infant rhesus Demo@-aPhic effects of home range dismonkeys and their mothers in a free-rang- Placement on rhesus monkeys at Chhatari,
ing band. ZOOLOGICA 51:17-27,1966.
Kumar, A.; Kurup, G.U. Infant development PRIMATOLOGY 10:433,1986.
in the lion-tailed macaque, Macaca silenus: Teas, J. BEHAVIORAL ECOLOGY OF RHEthe first eight weeks. PRIMATES 22512- sus MONKEYS (MACACA MULATTA)IN
KATHMANDU, NEPAL. PhD dissertation,
522, 1981.
Lee, P.C. Early infant development and ma- The Johns Hopkins University, 1978.
ternal care in free ranging vervet monkeys. Teas, J.; Richie, T.L.; Taylor, H.G.; Siddiqi,
M.F.; Southwick, C.H. Natural regulation
PRIMATES 2 5 3 6 4 7 , 1984.
Mack, D. Growth and development of infant of rhesus populations in Kathmandu, Nered howling monkeys (Alouatta seniculus) pal. FOLIA PRIMATOLOGICA 35:117-123,
in a free-ranging population. Pp 127-136 1981.
Без категории
Размер файла
1 185 Кб
development, infant, rhesus, constraint, independence, ecological
Пожаловаться на содержимое документа