PRIMATES, 36(2): 213--225, April 1995

213

Play Behaviour and Its Effects on Social Development of Common Chimpanzees (Pan troglodytes) NICOLA MARKUS and DAVID B. CROFT University of New South Wales

ABSTRACT. The aim of the study was to monitor the social development of infant and juvenile common chimpanzees (Pan troglodytes) through their play behaviour at Taronga Zoo in Sydney in order to examine the possible effects of captivity (such as limited opportunities to play) on social development. Play behaviour was observed by focal animal sampling to determine individual differences, and their relationship to age, sex, and relatedness of the subjects. Analysis revealed marked individual variations in social, solitary, and object play behaviours indicative of a relatively wellbalanced social and physical environment. Subjects showed a marked preference for play-partners of a different age compared to their own, and initiated interactions with similar frequency with members of both sexes. Many social-play dyads consisted of related individuals, and familiarity with prospective play-mates was the most decisive factor in social interactions. Key Words: Pan troglodytes; Play behaviour; Social development; Individual differences; Captivity.

INTRODUCTION Play is a vital part of primate development which helps to prepare a juvenile for its future role in society. The importance o f play only becomes apparent in its absence. Deprivation, particularly of social play, causes severe impairment of social development, including ineffective mating strategies, poor motor-pattern development and low motivation (LANCASTER, 1972). Inevitably, captivity places constraints on social development. Resources for play are limited and demands which encourage skill acquisition in the wild (e.g. efficient foraging) are not faced. Captivity eliminates survival pressures caused by predators, food competitors, environmental variables such as droughts or floods and the need to forage for food or to capture prey. Natural elimination of the 'least fit' is often overruled by the intervention of veterinary science. Captive breeding programs for the common chimpanzee (Pan troglodytes) have successfully ensured their survival and the challenge now is to conserve their behavioural repertoire. Emphasis must therefore be placed on assessment of the adequacy of social and physical conditions, and on mimicking as closely as possible the natural surroundings o f the animals. This study examined the following hypotheses about play in chimpanzees: (1) The frequency and duration of play declines with increasing age in juvenile chimpanzees (GOODALL, 1983; PUSEY, 1990). (2) Male infant/juvenile play is more aggressive than female play in preparation for adult reproductive competition, and males engage in more rough-and-tumble play than females (GOODALL, 1971; JOLLY, 1985; LANCASTER, 1972; OWENS, 1975), and play longer and more intensively than their female contemporaries (MCKENNA, 1982). (3) Chimpanzee infants/juveniles most frequently play with individuals of a different age/sex (SuGIYAMA, 1972), unlike vervet monkeys and most other primate

214

N. MARKUS~r D. B. CROFT

species (CHENEY, 1978; LEE, 1982). (4) A high frequency, complexity, and long duration o f play is an index o f adequate environmental richness for captive individuals (RosE, 1992). The diversity of play behaviours displayed at various ages indicates any apparent deficiencies in the physical or social environment. METHODS HOUSING The Taronga Zoo chimpanzee colony consisted of 27 members (2 adult males in isolation) of which 25 were observed: 1 adolescent male, 13 adult females, 7 infants (suckling), and 5 juveniles (weaned, <7 yrs). The troop contained six lineages, five of which were headed by an old female and further consisted of her children and grandchildren. In addition to these families, 'Snowy', an adolescent male of new bloodstock, was introduced to the group in 1983. Management of the colony includes frequent transfers of individuals to maximize genetic variation in this and other captive colonies. The colony was housed in a I-ha outdoor enclosure attached to a night-house. While outside, the chimpanzees occupied a large area consisting of grass and natural rock-platforms, surrounded by walls and a moat. Other features were a creek, palm trees, a large climbingframe of dead logs, and two artificial termite-mounds filled with jam to encourage tool-use in the troop. Climbing and swinging were encouraged by ropes suspended between the trees and the climbing frame. Data were only collected in the outdoor enclosure, where play and other social activity took place in surroundings most closely resembling the natural environment of the chimpanzees. A preliminary observation period was used to determine the optimum times of day for the recording of play, to identify the members of the troop and their relationships with each other and to establish and define categories of play. Play was most frequent before midday and after 13:45, as the chimps usually remained inactive in the midday sun. During the day, the colony was fed twice, at 10:30 and at 13:30, in addition to an early morning and an afternoon feed inside the night-house. Little play occurred during feeding, although a general increase in activity was noted immediately after each feed. Little or no play occurred during periods of rain or strong wind, and such days were avoided as observational periods. PLAY BEHAVIOUR Play behaviour was divided into three categories: solitary play, social play, and object play according to FAGEN (1981). Each category comprised a number of behavioural patterns which have previously been defined (GOODALL, 1968, 1971, 1973; MCGREW, 1977; NICOLSON, 1977). A brief summary of the units used in this study follows. Seven patterns of social play were used. (1 - 3) Three grades (gentle, average, and extreme) of rough-and-tumble (r-t). The behaviour consists of all kinds of wrestling behaviour; such as mutual grappling, kicking, grabbing with hands, hitting, biting, and rolling around on the ground. The gentle grade involved mild stimulation and manipulation of another individual and the extreme grade involved strong physical force with the potential to escalate into serious fighting. (4) Tickling was the physical stroking of an individual's skin with one or more fingers to usually elicit a squirming and wriggling response. (5) Chasing was the

Play in Captive Chimpanzees

215

pursuit on foot of an individual by one or more others. (6) Play-mothering was the nursing, cradling, and close observation of small infants by older juveniles. (7) Other social play was any other kind of play between two or more juveniles, identified either by context or facial expression ('play-face'). This included mutual play on the climbing frame and any tactile contact between individuals which was not identified as grooming. Eight patterns of solitary play were used. (1) Pirouetting involved spontaneous 360 ~ turns while either walking or stationary. (2) Handstand was the flipping onto the hands and either falling over or remaining in a handstand by leaning upside-down against a vertical support such as a wall or window. (3) Tumble-turn involved spontaneous forward or backward roils while moving or stationary. (4) Cartwheel was a 360 ~ turn through the air, landing first on one hand, then transferring body-weight onto the other hand, and finally landing upright in a standing position again. (5) Swing/Dangle involved hanging by two or four limbs from a rope, a branch, or under the abdomen of another individual. Swinging graded from gentle to vigorous. (6) Climbing was the process of moving at some distance above the ground in trees or in the climbing frame. (7) Blindman's Buff (DE WAAL, 1989) was a game which involved the self-handicapping of the player by placing some object (in this study a coconut-half) over its eyes to render it blind as it progressed through its environment. (8) Other solitary play was any other kind of non-social play which did not involve an object (e.g. young infants rolling around on their backs, with all four limbs suspended in mid-air while displaying a broad play-face). Six patterns of object play were used. (1) Hessian bag play, (2) coconut play, (3) bamboo play, and (4) other object play were kinds of play which involved the manipulation and mouthing of specific objects as they were available and did not include instances or objects where the object was consumed as food. Objects were also sometimes used to display at other individuals by waving them about vigorously but this was not scored as play since it meets the definition for quasi-aggressive behaviour (ADANG, 1986). (5) Jam-fishing made use of a long piece of bamboo which was first prepared by stripping it of all its side-branches. This was poked into one of several small holes in the artificial termite mound in order to extract the jam or honey. Jam-fishing begins as imitatory object play in infants, but becomes a goal-oriented activity as older juveniles become more proficient at it. (6) Kikuyu play involved only manipulation of the snake-grass by infants, but was gradually shaped into nest-making skills in later development. The remaining behaviour of the subjects was placed in one of two non-play patterns. (1) Other social was any kind of social, non-play activity between two or more chimpanzees such as allo-grooming, suckling, riding on a female's back, or sleeping in her embrace. (2) Other solitary was all other non-play behaviour, including feeding, self-grooming, walking, and complete inactivity. DATA COLLECTION AND ANALYSIS

Focal animal sampling was used to collect behavioural data. Each of the 12 infants/ juveniles (Table 1) was observed for a total of six one-hour periods, each taken on a separate day (June - September 1992). Observations were recorded using a dictaphone so that visual contact with the subject was maintained at all times. During social interactions, the partner(s) of the focal animal was recorded. Behaviour records were transcribed from the audio-tapes using Visual Basic programs created by the authors. First the sequence (and any amendments) were transcribed into the computer. Then the sequence was replayed with the tape to record the time of onset

216

N. MARKUS~z O. B. CROFT

Table 1. Age-groupings of chimpanzee subjects according to physical and social development. Age-group

Name

Sex

Age (months)

1: 2: 2: 2: 2: 3: 3: 3: 4: 4: 4: 4:

Kuma Sumu KiRe Chunga Mali Kibale Lobo Gombe Shona Lewis Monte Cheena

Female Female Female Female Male Male Male Male Female Male Male Female

9 15 16 17 18 28 38 44 58 80 82 84

Infant Infant Infant Infant Infant Infant Weaning infant Juvenile Juvenile Juvenile Juvenile Juvenile

and offset of each activity. Subsequent analysis resolved the frequency and duration of each activity and the matrices of different social interaction types between individuals. Due to the uneven age distribution of the sexes, data were primarily analyzed by individual rather than by sex and/or age class. Other studies (GOODALL, 1983; HAYAICI, 1988; PUSEY, 1990) have used two age classes (infants: 0 - 4 yr; juveniles: 5 - 7 yr) to divide the age range of subjects observed in this study. This classification has been followed using weaning to divide infants and juveniles but additionally four age groups ( < 1 yr, 1 - 2 yr, 2 - 4 yr, and 4 - 7 yr) were distinguished (Table 1). Interactions involving each unit of social play were analyzed separately to determine play-partner preferences. Extreme rough-and-tumble was too infrequent to warrant a separate behavioural unit, and so data were pooled with average r-t for analysis. A modified Chi-squared analysis (FAGErq & MANKOVICH, 1980) showed those dyads who engaged in more social play interactions of a particular type than expected if the initiator directed its social play randomly to all other individuals in the colony. Thus the results are based on the distribution o f frequencies of social play between individuals and not on the mean frequency of social play (see Fig. lb). A series of analyses compared the observed frequencies of dyads which had social play interactions significantly more often than chance against expected frequencies derived from the proportions of possible dyads in the colony. The frequency and duration of play was analyzed by category, i.e. social, solitary, and object play. However, Blindman's Buff which had both a social and solitary component was included in both social and solitary play. To determine any apparent clustering of individuals on a combination o f behaviours within the main categories of play, a factor analysis (SPSS/PC + Version 4.0) was performed using either the mean frequency or the duration of the units comprising a category.

RESULTS AGE-RELATED TRENDS IN PLAY

Solitary play increased sharply up to 30-months of age, and then decreased to almost complete cessation at 84 months (Fig. la). Sharp dips at 38 months represent a young male, Lobo, whose characteristically active play-patterns observed during the preliminary study were suppressed by the weaning process.

Play in Captive Chimpanzees

217

70-

~50

60-40 50- 30

4030-

20

20-10

100

O

2O

2O

=10-

I0~

|

0

o

20

30 (c) -20

-10

o

o

i

i

,o

~o

;o

9

,;

~o

~'o

;0'8'o

D

911

Age (months)

Fig. 1. Mean (+ l SE) frequency per hour (iN female; 9 male) and mean (+ 1 SE) percent time (9 female; 9 male) of (a) solitary play (b) social play, and (c) object play with age of subject (n=6 observation sessions).

Social play peaked slightly earlier than solitary play at about 16-months of age, and then decreased at a considerably slower rate (Fig. lb). Overall, social play occurred substantially less often and took up less time than solitary play at any age. A sharp drop at 38 months again represents Lobo. Among the oldest juveniles, social play decreased rapidly in frequency but less so in duration, indicating fewer but slightly longer play-bouts. Changes in object play with age were less conclusive than in the other two play categories (Fig. lc). Initial decreases in both frequency and duration of object play were countered by a sharp rise at 17 months of age. In addition, contrary to trends in both solitary and social play, Lobo engaged frequently in object play. The oldest age group showed conflicting trends in the frequency and duration of object play. One exhibited the low frequency of object play-bouts for his age, yet he simultaneously displayed the most time spent in object-play of any age. The other had a high frequency but short duration of object play. PLAY PARTNER SELECTION

Significant positive social play interactions were found within all infant and juvenile classes and between these classes except for male infants and female juveniles (Table 2). Such interactions involving adult females were confined to infants under 18 months. The single non-zero value for the male, Snowy, was play initiated only by Shona, a female of the oldest juvenile group.

218

N. MARKUS• D. B. CROFT

Table 2. Number of significant positive social-play interactions per dyad between male infants (MI), female infants (FI), male juveniles (M J), female juveniles (FJ), female adults (FA), and the young male Snowy (S). To:

MI

From: MI FI MJ FJ

0.67 0.33 0.67 0.00

FI (3) (12) (9) (6)

0.42 0.33 0.83 0.88

MJ (12) (6) (12) (8)

0.67 0.67 0.33 0.50

FJ (9) (12) (3) (6)

0.00 0.88 0.67 1.00

FA (6) (8) (6) (1)

FA 0.03 (36) 0.06 (48) 0.00 (36) 0.00 (24) S 0.00 ( 3 ) 0.00 ( 4 ) 0.00 ( 3 ) 0.00 ( 2 ) Numbers in brackets indicate the total number of possible dyads.

0.03 0.04 0.00 0.00

S (36) (48) (36) (24)

0.00 0.00 0.00 1.00

(3) (4) (3) (2)

0.00 (66)

0.00 (12)

0.00 (12)

--

(0)

When interactions among infants and juveniles were examined, no apparent preferences for play-partners of the same sex compared to partners of the opposite sex were found, irrespective of the sex of the initiator (xz= 0.29, df= 2, n.s.). Individuals strongly preferred play-partners two age-groups removed from their own (X2= 14.0, df= 3, p < 0.01). Twentytwo of these inter-age-group interactions included members of the same family, who were either sisters, or uncles and nieces/nephews. Dyads consisting of related individuals therefore comprised 33~ of all significant interactions, significantly more than expected by chance (X2=15.3, df=3, p<0.001). The hypothesis that young males play in more aggressive relationships than females was not supported by the data. Five out of eight juvenile dyads contained both sexes, two contained only males and one contained only females. Out of ten significant positive roughand-tumble dyads, five involved play initiated by males, four by females, and one by an adult. A similar result was obtained in the 'chase' category. Seven of 13 chases were initiated by males, and six by females. Equal numbers of dyads were uni-sexual and bi-sexual, and one interaction was directed at a female. However, of six uni-sexual relationships only two contained females. A larger sex bias was observed in gentle rough-and-tumble play. O f 12 play-bouts initiated by infants/juveniles, 9 were initiated by females and only 3 by males. A possible explanation for this may be the large involvement of infants, who were predominantly female. Males participated equally as much as females in traditionally female, sex-role priming behaviour. Play-mothering/tickling yielded seven significant positive interactions, of which four were initiated by males. However, this may again be explained by the age/sex-bias in the group, where five males were at a potential play-mothering age as compared to only two females. INDIVIDUAL DIFFERENCES IN PLAY

The frequency of solitary play was mainly related to age (Fig. 2a). The lower corner of the plot represents the members of the oldest age-group who rated low on all aspects of solitary behaviours. The extension of this grouping towards the right corner of the plot shows the two younger age-groups, who generally engaged more often in solitary play such as acrobatics, tumbling, and 'other solitary' behaviour. Notable exceptions to any clustering were two individuals on the left side of the plot. A high incidence of acrobatics, combined with the only display of Blindman's Buff in the group, represented (8) Gombe, a young male of unique acrobatic ability. At the opposite end of the scale, rating highly only on

Play in Captive Chimpanzees

219

, ~:i 9 ! :i:i-i 84 i

i i 84 ! i

: ,

~ 84.....

~~

i !i84184 i!:

20txerAUdiencesolitaryplayplay

:

0

Factor 3

z 1.5

1

F

"

for2

-

itor,

~

05

-1

-1 b

1.s

1.5

0.5

o.s

-0.5

-0.5

Factor 3

- I .5 3,5

-I .S

Other solitary play Audience play

1 .g

1

1.5 2

-0.6

" ~ - o . 6

Tumbling

Fgemr 1

-1.5

Fig. 2. (a) Factor scores for mean frequency per hour of solitary play and other solitary behaviour on factors 1 - 3, which explain 48.1, 23.5, and 13.4% of the variance, respectively. (b) Factor scores for mean % time (arcsine and square root transformed) spent in solitary play and other solitary behaviour for factors 1 - 3 , which explain 35.4, 27.6, and 17.6% of the variance, respectively. Axes are labelled according to the principal factor loading of the behaviour unit. Numbers indicate individual chimpanzees in order of ascending age. 1: Kuma (f); 2: Sumu (f); 3: Kike (f); 4: Chunga (f); 5: Mall (m); 6: Kibale (m); 7: Lobo (m); 8: Gombe (m); 9: Shona (f); 10: Lewis (m); 11: Monte (m); 12: Cheena (f).

audience play and 'other solitary' play is (1) K u m a , the youngest infant whose play at only 9 months of age was still tentative. Duration o f solitary play showed a slightly different distribution o f individuals, although age groupings remained similar (Fig. 2b). The oldest age-group devoted little time to solitary play and engaged more in other solitary activities. The majority of younger infants around the centre of the plot distributed their time relatively equally among various solitary play activities. Two closely associated points at the centre of the plot represent two young males (5 & 6) who, although adjacent in age, are not in the same developmental age-group. This association indicates, that rates of development may vary in different aspects of play. Again, the two lone chimpanzees situated apart from the others show the youngest (1) and the most acrobatic (8) individuals, who spent the most time in non-specific solitary play and highly active play, respectively. Social play plots reveal a wide distribution of chimpanzees in both frequency (Fig. 3a) and duration (Fig. 3b) of play. A separation of the subjects on the frequency plot occurred along the diagonal of factors 1 and 2. The left half of the plot contains the two older agegroups, most of w h o m demonstrated high frequencies of rough-and-tumble and chasing, but moderate to little other social activity. The right side o f the plot represents individuals

220

N. MARKUS~: D. B. CROFT

i

: 0

~

!84184184 i

i ~ i i /

" :

7

'

:

1

:

I:

....1

-o~ o5~

2

:

, 9 0 Factor3

"

-1

" .

~

Factor 2

3 . :

"

-Z 1.5

14~

i:: 84. !

•

9

_

1

Other social play

]

Other social activity

)

~ -1.5

b

1.5 o .

5

~

: s

: ' : 2:

~I

, -o 5

] Other:&t ]

'

.....

'6

.

~5

0~

~

.

:: .

[ 00 . 5 "9I~ [ ~ .l~__

r ,oymo oo, r

" "'oB1+5i

~

t

,.o,o.,

" "

[ ~)~netlreroci&:iplay I

-I .5

Fig. 3. (a) Factor scores for mean frequency per hour of social play and other social behaviour on factors 1 - 3 , which explain 30.5, 28.2, and 18.5~ of the variance, respectively. (b) Factor scores for mean 070 time (arcsine and square root transformed) spent in solitary play and other solitary behaviour for factors 1 - 3 , which explain 32.8, 27.2, and 17.0070 of the variance, respectively. See Figure 2 for explanation of labelling.

who rated high on 'other social' play and non-play activity, but low on the more active social play behaviours. The two oldest juveniles (11 & 12) rated low on all social play behaviours. Frequencies of play-mothering and gentle rough-and-tumbling were low, and only two infants (3 & 4) and the juvenile female (9) Shona were strongly involved in these. Duration of social play showed an obvious separation of two young infants (2 and 3) to the right of the plot (Fig. 3b). This position indicates high durations of tickling, gentle rough-and-tumbling and 'other social' play, but extremely low ratings on chasing, roughand-tumble and 'other social' activity. Only the youngest infant (1) Kuma rated high on 'other social' activity and is therefore located above the others on the plot. The young female (9) Shona showed notably high durations of play-mothering which coincided with her high frequency rating on this behaviour. By contrast, all members of the second oldest age-group rated low on durations of play-mothering and 'other social' activity, but engaged in longer periods of chasing and rough-and-tumbling. Frequencies of object play yielded the most obvious separation of subjects of all threedimensional analyses (Fig. 4a). Three of the oldest juveniles form a cluster in the lower corner of the plot and rated high on jam-fishing and low on most other object-play. Two exceptions were (11) Monte, who rated high on most kinds of object-play, and (8) Gombe, whose game of Blindman's Buff again separated him from the other troop members. Two other individuals relatively closely associated on the right side o f the plot were the youngest

Play in Captive Chimpanzees

221

other obj~ Bamboo p] -><. -1.5

tl

2 1.S

6

Factor 2

o

:;,o

-O.S

9

d

;

s

-1

-1.G

-1

~.S

0

Factor I

0.5

1

I.S

2

2.6

Bamboo play Hessian play Other object play

I

I

Fig. 4. (a) Factor scores for mean frequency per hour of object play on factors 1, 2, and 3, which explain 39.2, 19.3, and 17.2% of the variance, respectively. (b) Factor scores for mean % time (arcsine and square root transformed) spent in object play for factors 1 and 2, which explain 33.2 and 22.8% of the variance, respectively. See Figure 2 for explanation of labelling.

infant (1) and the 3-yr-old male, (7) Lobo. Both rated moderately on kikuyu, hessian, and coconut play and low on all other object play. The most interesting result was the clear separation between individuals engaged in jam-fishing (9, 10, & 12) and those using bamboo for other kinds of play ( 2 - 5). This difference suggests a progression from manipulatory object play during infancy to the skilled usage of bamboo in the later juvenile stages. However, although most older chimpanzees abandoned object play for goal-directed jam-fishing behaviour, one individual (11) demonstrated that both need not be mutually exclusive. The duration of object play (Fig. 4b) revealed a cluster of chimpanzees at the centre of the plot. From all age groups, these individuals showed moderate durations of object play. However, there were four individuals outside this cluster. The young male (8) G o m b e was again isolated by his game of Blindman's Buff. G o m b e also shared high durations of coconut and kikuyu play, but no rating on jam-fishing, with the youngest infant (1) Kuma. By contrast, the oldest juvenile (11) M o n t e displayed exceptionally high durations of jamfishing, but low ratings on all other object play. Finally, the infant (3) Kike showed a very high rating on durations of bamboo, hessian, and other object play compared to her age-mates (2, 4, and 5), who had all displayed high frequencies of this behaviour (Fig. 4a), but all with the exception of Kike rated low on durations of object play. This indicated the generally short attention span of young infants and of many older individuals in object play.

222

N. MARKUS& O. B. CROFT

DISCUSSION The frequency and duration o f play peaked early on in development and decreased with increasing age as in previous research (LEE, 1982) and in populations of wild chimpanzees (GOODALL, 1971; HAYAKI, 1985; PUSEV, 1990). Various units of play behaviour peaked at different ages in accordance with young chimpanzees' increasing ability to interact with their environment. However, whereas trends were easily detectable in solitary and in social play (despite some obvious exceptions), object play appeared to depend on some less predictable factor than mere age. The observed digressions from expected trends in play demonstrated the individualistic responses of chimpanzees to a given environment. Lobo, a young male who had been one of the most active participants in play during the preliminary observation period, was at 38 months fast approaching the weaning age and therefore the end of his infancy. The trauma of weaning has been previously recorded in wild chimpanzees, and is known to cause various degrees of depression and stress to individual infants (GooDALL, 1990). Lobo's mother's increasing rejection of his advances to suckle resulted in several weaning tantrums and long periods of complete inactivity. This drastic change occurred almost overnight and the infant had not resumed playing by the end of the study, which naturally affected his play-ratings. A less predictable incongruity in play-patterns was caused by Monte, who at 82 months displayed consistently greater amounts of play than his age mates. Unlike the other three chimpanzees in his age-group, Monte's routine often included the performance of spontaneous pirouettes while walking. This pirouetting accounted for his relatively high rating on solitary play which was matched only by the much younger and extremely acrobatic Gombe. Although excessive pirouetting is often observed in severely under-stimulated captives (GoODALL, pers. comm. with Lolzos, 1967), Monte's generally high amounts of play seem to indicate that this is not the case. Furthermore, Monte's apparent fascination with all young infants produced a consistent rate of play-mothering. Examination of Monte's preoccupation with infants showed that most was the result of the swift abduction and detention of infants against their will. Relatively little participation in other social play, but a much greater amount of object play than expected, seem to indicate a lack of appropriate challenges for a young male of Monte's age. Analysis of social activity provided results in line with expectations. Play-partners of the opposite sex were chosen equally as often as same-sex partners, despite the ample availability of juveniles of both sexes. In addition, a strong preference for partners two age-groups removed was observed in all categories of social play. Both results showed, that in captivity as in the wild, interaction of age mates is infrequent, regardless of their availability (SuGIVAMA, 1972). The apparent attraction of different age-groups involved a large number of related dyads, which suggests the importance of familiarity with prospective play-partners. The preference for play-partners two age-groups removed supported this idea, as related individuals were almost always separated by two age classes. Nevertheless, about 60% of unmatched dyads were not related. This outcome suggests, that there is a natural fascination for unmatched play-mates which is most obviously expressed in colonies with a wide range o f juveniles. This contrasts with the wild where chimpanzee mothers with infants are less sociable than males or oestrous females and tend to isolate themselves in a small core area with their infants and juveniles (WRANGHAM• SMUTS, 1980). Accordingly, their offspring have few opportunities to play with individuals other than their siblings or mother (HAYAKI, 1988; PUSEV, 1990). Thus in captivity infants and juveniles

Play in Captive Chimpanzees

223

potentially have a wide range of play-partners who are continuously available but even so our results show some preference for close relatives o f inevitably different ages as would be found through the constraints of the mother's behaviour in the wild. The ability of a mother and her pre-adolescent offspring to form a discrete unit in a dense captive population should therefore be investigated. A large age-range in the case of Taronga's chimpanzees placed a number o f individuals at an age between infancy (i.e. above 3 yrs) and adolescence (at more than 7 yrs of age), and thereby at the bottom of the hierarchy. Consequently, all younger infants, who were still under the protection of their maternal ranking, were themselves ranked above the juveniles to various degrees. The preference for high-ranking play-partners has been variously documented (CHENEY, 1978; LEE, 1982). However, the reciprocated selection of older play-mates by younger infants suggests a mutual curiosity, which may be fuelled by the initial interest shown by the older juveniles. Regardless of its origins, the establishment o f familiarity and trust among group members favours the socialization of all infants and juveniles and is thereby mutually advantageous. Differences in the vigour of play between male and female infants proved impossible to establish conclusively. Pooled data of extreme and average rough-and-tumble showed a minimally higher rate of male over female initiation of play-bouts. Similar results were obtained for chasing, although this behaviour showed a slightly higher preference for samesex play-partners in males. Factor analysis of social play showed all females to rate low on frequency and duration of rough-and-tumble and chase behaviours. Both categories, however, involved infants and juveniles alike, and as almost all infants happened to be female, the outcome seems to provide a more conclusive result in favour of age-preferences rather than sex-preferences. Similarities in aggressive play between male and female juveniles were previously observed in vervet monkeys (LEE, 1982). Therefore, although there was a slight trend towards more aggressive play in males, further research is required to obtain sufficient data to confirm this. Play-mothering and tickling showed a greater overall number of significant initiations by males. However, the highest ratings on both frequency and duration of the two behaviours were shown by the young female Shona. Data may therefore be indicative of the lack of females in the older age-brackets. This bias was exacerbated by the relative lack of involvement of Shona's age-mate Cheena, whose participation in social activity was almost completely limited to her younger sister Chunga. Two explanations for these gender-related behaviours have previously been given. Firstly, juveniles who are expected to play-mother are those who have the greatest reproductive costs as adults (FEDIGAN,1972). The acquisition of good mothering skills would be advantageous to the propagation of genes and would therefore be most important to female chimpanzees. Alternatively, infants may become strategic pawns in a process of rank achievement as in barbary macaques (JOLLY, 1985). The results of this study favour neither one of these hypotheses. Apart from the greater number of males to play-mother and tickle infants, some of the initiators were not in positions of concern with rank-achievement. In one case this was due to the young age o f the individual at 28 months, in another case the juvenile's close association with his high-ranking mother still assured him ample advantages over lower ranks despite his own low ranking. In addition, two other tickling bouts each involved two individuals who were closely related and who were thereby unable to influence relationships with high-ranking individuals outside their own families. Several approaches to the assessment of well-being of captive populations include behavioural and physiological tests for the presence o f abnormal behaviours, effects o f

224

N. MARKUS& D. B. CROFT

environment on development and comparisons with animals in known stressful conditions. Amendments to the U.S. Animal Welfare Act of 1985 consider the importance of providing " a physical environment adequate to promote the well-being" specifically for primates (ROSE, 1992). To assess such well-being, NOVAK and SUOMI (1989) proposed four criteria, which maintain that an animal should be in good physical health, exhibit a substantial range of its species behavioural repertoire, respond effectively to environmental changes and not be in a state of chronic distress (ROSE, 1992). Although the authors of these criteria believe that any two would suffice to guarantee well-being in captivity, in view of the ultimate aim to maintain releasable colonies, it is uncertain whether all the four criteria would suffice to produce individuals capable of surviving in the wild. Play behaviour observed at the Taronga Zoo colony was both plentiful and diverse, and chimpanzees were seen to utilize every imaginable aspect of their environment. In terms of NOVAK and SUOMI'S criteria, individuals easily met all of the requirements for well-being. A p a r t from features provided for the specific purpose of play, individuals were observed playing with bottle-tops, food wrappers, plastic pieces, paper, string, and a dead bird, most of which had been contributed by spectators. Hessian bags, coconut shells, b a m b o o sticks, m a c a d a m i a nuts, straw, strands of kikuyu, fern, and palm fronds were all utilized for various solitary and social 'games.' Nevertheless, due to the large number of juveniles, there were considerable differences in play-behaviours exhibited by individual chimpanzees. Gombe's acrobatic skill and innovative play was unique amongst the group. As the only individual to play Blindman's Buff (DE WAAL, 1989), Gombe's development of acute skills o f distance judgement while climbing, swinging, and jumping were frequently observable and unequalled. His ability to play innovatively indicates sufficient mental stimulation to prevent boredom and to encourage healthy development in a social and physical sense. However, individuals above 6 yrs of age displayed fewer interactions and the least activity of all subjects. They often were completely inactive. At an age where maternal dissociation is complete yet responsibilities of parenthood are absent, adolescents appeared least challenged by their environment. One individual at 82 months of age exhibited unusual amounts of object play and spontaneous pirouetting, and a preoccupation with infants which often caused him to be reprimanded by older females. Such behaviour may represent attempts to gain attention and to avoid boredom, and social development at this stage should be monitored to ensure further healthy growth in a social context. Problems of this difficult stage of development are exacerbated in captivity by the lack of interaction with other chimpanzee troops and the enforced incarceration of the usually highly mobile individuals (PUSEY, 1990). The absence of adult males in the study group further reduced the variety of individuals available for potential interactions. Such interactions are an important aspect of the development of social relationships in adolescent males and pubescent females in the wild (HAYAKI,1988; PUSEY, 1990). As the exchange of individuals is a vital aspect of chimpanzee management, a possible means of avoiding stagnation of social development may be the exchange of juveniles on reaching adolescence in a way which matches the well-documented change and development of social relationships of adolescent chimpanzees in the wild (GoODALL, 1986; HAYAKI, 1988; PUSEY, 1983, 1990).

Acknowledgements. Permission to undertake the study at Taronga Zoo was granted by the Director of Scientific Policy and Research, Dr. J. GILES,through the Zoological Parks Board Ethics Committee. We thank Dr. GILESand the Zoo's primate keepers, especially DEBBIECOX,for their encouragement and assistance with the study. PETER O'DoNOVANassisted in the practicalities of the study. We thank Prof. D. SA~EMAN, Dr. W. SHERWtNfor constructive comments on earlier drafts of this paper.

Play in Captive Chimpanzees

225

REFERENCES ADANG, O. M. J., 1986. Social development of chimpanzees: the role of social exploration. In: The Individual and Society, L. PASSERA t~r J. P. LACHAUD(eds.), Private, I.E.C. Toulouse, pp. 101 - 108. CHENEY, D. L., 1978. Play partners of immature baboons, Anita. Behav., 26: 1038- 1050. FAGEN, R. M., 1981. Animal Play Behaviour. Oxford Univ. Press, Oxford. - ~r N. J. MANKOVICH, 1980. Two-act transitions, partitioned contingency tables, and the 'significant cells' problem. Anita. Behav., 28: 1017-1023. FEDIGAN, L., 1972. Social and solitary play in a colony of vervet monkeys (Cercopithecus aethiops). Primates, 13: 3 4 7 - 364. GOODALL, J., 1983. Population dynamics during a 15 year period in one community of free-living chimpanzees in the Gombe National Park, Tanzania. Z. Tierpsychol., 6 1 : 1 - 6 0 . - - , 1986. The Chimpanzees of Gombe. Harvard Univ. Press, Cambridge, Massachusetts. - - , 1990. Through A Window. Weidenfeld & Nicholson, London. GOODALL, J. VAN LAWICK, 1968. The behaviour of free-living chimpanzees in the Gombe Stream Reserve. Anita. Behav. Monogr., 1:161-311. - - , 1971. In the Shadow of Man. Collins, New York. - - , 1973. Cultural elements in a chimpanzee community. In: Precultural Primate Behaviour, E. W. MENZEL (ed.), Karger, Basel, pp. 144-184. HAYAKI, H., 1985. Social play of juvenile and adolescent chimpanzees in the Mahale Mountains National Park, Tanzania. Primates, 26: 3 4 3 - 360. - - , 1988. Association partners of young chimpanzees in the Mahale Mountains National Park, Tanzania. Primates, 29: 147-161. JOLLY, A., 1985. The Evolution of Primate Behaviour (2nd ed.). MacMillan, New York. LANCASTER, J. B., 1972. Play-mothering: the relations between juvenile females and young infants among free-ranging vervet monkeys. In: Primate Socialization, E E. PO1RIER (ed.), Random House, New York, pp. 83-105. LEE, P. C., 1982. Play as a means for developing relationships. In: Primate Social Relationships, R. A. HINDE (ed.), Blackwell Scientific, London, pp. 225-229. Loizos, C., 1967. Play behaviour in higher primates: a review. In: Primate Ethology, D. MORRIS (ed.), Weidenfeld & Nicholson, London, pp. 176-215. McGREw, W. C., 1977. Socialization and object manipulation of wild chimpanzees. In: Primate Bio-social Development, S. CHEVALIER-SKOLNIKOFFt~ E E. POIRIER (eds.), Garland, New York, pp. 261-288. MCKENNA, J. J., 1982. The evolution of primate societies, reproduction and parenting. In: Primate Behaviour, J. L. FOBES & J. E. KiNG (eds.), Academic Press, New York, pp. 123-164. NICOLSON, N. A., 1977. A comparison of early behavioural development in wild and captive chimpanzees. In: Primate Bio-social Development, S. CHEVALIER-SKOLNIKOFF• E E. POIRIER (eds.), Garland, New York, pp. 529-560. NOVAK, M. A. & S. J. SUOM1, 1989. Psychological well-being of primates in captivity. ILAR News, 31: 5 - 1 5 . OWENS, N. W., 1975. Social play in free-living baboons, Papio anubis. Anita. Behav., 2 3 : 3 8 7 - 4 0 8 . PUSEY, A. E., 1983. Mother-offspring relationships in chimpanzees after weaning. Anita. Behav., 31: 363 - 377. - - , 1990. Behavioural changes at adolescence in chimpanzees. Behaviour, 115: 2 0 3 - 246. RosE, M. A., 1992. Monkey housing: a room, a tree and a puzzle. ACCART News, 5: 3 - 5. SUGIYAMA, Y., 1972. Social characteristics and socialization of wild chimpanzees. In: Primate Socialization, E E. POIRIER (ed.), Random House, New York, pp. 145-163. DE WAAL, E B. M., 1989. Peacemaking Among Primates. Harvard Univ. Press, Cambridge, Massachusetts. WRANGHAM, R. W. & B. B. SMUTS, 1980. Sex differences in behavioral ecology of chimpanzees in the Gombe National Park, Tanzania. J. Reprod. Fertil. (Supplement), 28: 1 3 - 31. --

Received: April 11, 1993; Accepted: October 17, 1994

Authors' Names and Address: NICOLAMARKUSand DAVIDB. CROFT,School of Biological Science, The University of New South Wales, Sydney, New South Wales2052, Australia. e-mail Address: [email protected].