Ethology

The Context and Quality of Social Relationships Affect Vigilance Behaviour in Wild Chimpanzees Nobuyuki Kutsukake Department of Cognitive and Behavioral Science, Graduate School of Arts and Sciences; and Department of Biological Sciences, Graduate School of Sciences, University of Tokyo, Tokyo, Japan

Correspondence Nobuyuki Kutsukake, Department of Biological Sciences, Graduate School of Sciences, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. E-mail: [email protected] & [email protected]

Received: January 24, 2005 Initial acceptance: March 30, 2005 Final acceptance: September 26, 2005 (K. Riebel) doi: 10.1111/j.1439-0310.2006.01200.x

Abstract Previous studies of vigilance behaviour have focused mainly on the influence of predation threat, whereas the influences of conspecific factors, such as within-group threats, are relatively unstudied. To elucidate the influences of conspecific factors, this study examined vigilance behaviour in wild chimpanzees (Pan troglodytes schweinfurthii) in Mahale Mountains National Park, Tanzania. Vigilance level was lower during foraging than during resting, which indicated a conflict between vigilance and foraging activity. In addition, vigilance level was higher when chimpanzees were on the ground where an encounter with leopards (Panthera pardus) is likely than when the chimpanzees were in trees. Males, but not females, increased their level of vigilance as the number of individuals within 3 m increased. In both males and females, daily party size – an index of group cohesion – did not affect the vigilance level. The level of maternal vigilance was higher when a dependent infant was separated from its mother than when the offspring was in contact with its mother. Both males and females increased their vigilance when a less-associated group member was nearby, when compared with when there was no less-associated group member nearby. This finding suggests that variation in relationship quality influences the vigilance level and that individuals need to increase their level of vigilance when the level of within-group threats is high. This study indicated that variation in vigilance cannot be understood unless conspecific factors, such as variation in the relationship quality with associates, are considered.

Introduction The main function of vigilance behaviour is commonly assumed to be a search for predation threats (reviewed in Bednekoff & Lima 1998; Lima & Bednekoff 1999a,b; Treves 2000; Caro 2005). Supporting this idea, ‘group size effect’ – a negative relationship between an individual vigilance level and group size during foraging activity – has been repeatedly confirmed in vigilance studies in mammals and birds (Elgar 1989). However, the presence of a group-size effect is debatable as relatively few studies supporting the group-size effect have controlled for potenEthology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

tially confounding factors that affect the individual vigilance level, e.g. group composition or quality of intra-group social relationships (Beauchamp 2001, 2003 and commentaries; Elgar 1989; Quenette 1990; Blumstein et al. 1999; Lima et al. 1999; Treves 2000; Caro 2005). In particular, recent studies highlight the importance of conspecific factors as a determinant of vigilance. For example, mammal and bird species increase their vigilance (conspecific monitoring) in the presence of a competitive conspecific (e.g. Knight & Knight 1986; Roberts 1988; Jones 1998; Goss-Custard et al. 1999; Robinette & Ha 2000; Barbosa 2002; Coolen 2002; Cameron & du Toit 2005). 581

Vigilance Behaviour in Chimpanzees

Part of the vigilance is directed toward group members (Caine & Marra 1988; Alberts 1994; Watts 1998; Treves 2000), particularly toward higher ranking individuals (Keverne et al. 1978). Negligence of conspecific influences on vigilance behaviour seen in the previous studies partly came from the fact that those studies usually observe unidentified individuals (e.g. Bertram 1980; Heathcote 1987; Hunter & Skinner 1998; see Treves 2000; Hirsch 2002). As a result, for instance, variation in the relationship quality with associates has been rarely investigated as a determinant of vigilance in social animals. Given that the level of conspecific threats and probability of the occurrence of competition should vary with whom the individual interacts (e.g. related vs. unrelated group members), vigilance should be influenced by the relationship quality with associates. For example, Watts (1998) showed that the monitoring rate of female mountain gorillas (Gorilla gorilla beringei) was higher when interacting with other females with whom they have hostile relationships than when interacting with familiar group members. In addition, previous studies have shown that maternal vigilance functions to monitor and protect the safety of dependent offspring (Caro 2005). Mothers showed elevated vigilance relative to non-mothers in mammals (Burger & Gochfeld 1994). Maternal vigilance increases after the birth of infants and when the probability of an attack on the infant from a predator or conspecific is high (Maestripieri 1993a; Treves 1999a; Treves et al. 2003) or infanticide is likely (Steenbeek et al. 1999). To better understand the factors that influence vigilance behaviour, researchers need to consider not only predation pressure but also various independent factors including social factors simultaneously. This study investigated vigilance behaviour in wild chimpanzees in Mahale Mountains National Park, Tanzania. Chimpanzees live in multimale, multifemale societies, in which group members associate in temporary parties of various sizes and compositions. Males are philopatric and are more gregarious than females (Goodall 1986). Chimpanzees are an interesting subject for studies of the conspecific influences of vigilance behaviour because individuals compete for dominance rank, food, and mates, and the relationship quality varies within a group (Goodall 1986; Kutsukake 2003). Furthermore, individuals use visual information as a cue for decision making during social interactions (Call 2001). First, I investigated how vigilance is influenced by the risk of predation. At Mahale, leopards Panthera pardus constitute the only potential chimpanzee 582

N. Kutsukake

predator at this site (see Boesch 1991; Zuberbuhler & Jenny 2002). Because leopards hunt chimpanzees on the ground, I predicted that (1) Chimpanzees are more vigilant when they are on the ground than when they are in trees; (2) ground-level vigilance decreases as the number of proximate individuals or the party size increases (van Schaik & van Noordwijk 1989; Rose & Fedigan 1995; Treves 1998, 1999c, 2000; Steenbeek et al. 1999; Treves et al. 2001). Conversely, vigilance does not necessarily decrease with increasing the party size or the number of proximate individuals when the individuals are free from predation pressure (i.e. are in a tree). In addition, I investigated how conspecific factors influence vigilance behaviour. Because conditions in which group members were in proximity create more possibilities for group member interactions, I predicted that (3) the vigilance level increases as the number of proximate individuals or the party size increases (Treves 1999a; Hirsch 2002). Note that this prediction is contrary to the prediction of the antipredator hypothesis. Concerning the variation in the relationship quality with associates, I predicted (4) vigilance level increases when a risky individual (e.g. a less associated or a dominant group member) is nearby. Finally, I investigated the factors affecting maternal vigilance. I predicted that (5) maternal vigilance should increase when an infant strays from its mother relative to when the infant was in contact with the mother. This is because the risk of predation or conspecific attack on an infant is high when the infant strays from its mother, and mothers need to check the position of their infants or to scan their surroundings. If chimpanzee mothers are cautious about the risk of predation, there should be a marked increase in maternal vigilance when the mother–infant pair is on the ground and no group members are nearby. By contrast, if the chimpanzee mothers are cautious about the risk of infanticide or an attack on the infant by a group member, the level of maternal vigilance should be higher when there are neighbour(s) nearby, irrespective of the height from the ground. Methods Study Site and Group

This study examined wild chimpanzees living in ‘M-group’ at the Mahale Mountains National Park on the eastern shore of Lake Tanganyika in western Tanzania. Researchers have been studying Mahale’s wild chimpanzees since 1965 (Nishida 1990) and all individuals, except for a few recently Ethology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

Vigilance Behaviour in Chimpanzees

N. Kutsukake

immigrated females, were well habituated. For this study, observations were made between Sep. and Dec. 2000 and between Feb. and Sep. 2001 (Kutsukake & Matsusaka 2002; Kutsukake 2003; Kutsukake & Castles 2004). During this time, the study group (M-group) ranged from 51 to 54 individuals, including eight adult males (>15 yr) and 20 adult females. In addition, one adolescent male (PM) that had reached adult body size and had received pant-grunts from adult females was included in the adult set. All nine adult males and nine adult females were selected as observation targets (Table 1). I chose the females so that the age distribution was not biased and represented the group composition. More importantly, I chose well-habituated females because a human presence influences the anxiety level and vigilance behaviour of shy females. As female reproductive condition may influence the vigilance behaviour, all observations were done when the focal females were not oestrous. All the focal females had a dependent offspring whose ages at the beginning of the observations ranged from 5 mo to 5 yr (x ¼ 2.2 yr). Of the nine focal females, one female (XT) gave birth soon after the start of Table 1: Identities of focal individuals and sample size No. of samples collected by the instantaneous sampling method

Male

ID

Birth

Dominance rank

Without neighbours

With at least one neighbour

DE MA FN HB DG BB AL CT PM

1963? 1977 1978? 1980? 1981? 1981 1985 1985? 1988

4 5 1 7 2 8 3 6 9

145 197 230 280 227 329 220 214 277

208 204 155 162 99 121 192 195 139

181 141 166 175 137 186 255 267 216

219 168 227 225 228 238 155 174 205

Offspring Female

FT OP PI JN XT MJ AK CY AB

1963? 1971? 1972? 1974? 1975? 1980? 1981? 1982? 1982

88M (PM), 99F 86F, 91M, 98M 91M, 00F 95F, 00F 95M, 00Fa 96M, 01Fa 98F 98M 98F

Offspring: ‘88M’ indicates that there was a male offspring born in 1988 within the group during the study period. athese infants were born during the study period. Ethology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

observations and one (MJ) gave birth later in the observation period (Table 1). For this mother, I used the data from before the birth. Observation Methods

I observed the 18 individuals using the focal animal sampling method (Altmann 1974). On each observation day, I chose one focal individual to follow for as long as possible; the choice of the focal individual was based on the criterion that data be accumulated evenly for every individual. I did not observe the same individual on two successive days (Kutsukake 2003). In total, I observed each focal individual for at least 50 h, and the total observation time was about 1100 h. An auditory encounter with leopards occurred ten times during those focal observation. Typically, chimpanzees showed a fearful reaction. That is, they climbed a tree and gave alarm calls. These reactions were identical to chimpanzee behaviour observed in a group in which predation by leopards was reported (Boesch 1991). In addition, the killing of a leopard cub by chimpanzees has been previously observed at this study site (Hiraiwa-Hasegawa et al. 1986). Although no evidence of predation by leopards was found at this study site, these episodes suggest that leopards are a threat to chimpanzees. During the focal observations, I recorded whether the focal individual was vigilant, any activity, the focal individual’s height from the ground, and the number and identities of group members within 3 m (hereafter ‘neighbours’) every 5 min using the instantaneous sampling method (Altmann 1974; van Schaik & van Noordwijk 1989; Smith et al. 2004). Previous studies on primates have used various definitions of vigilance behaviour. For example, some studies have defined movement of the head as vigilance (Steenbeek et al. 1999), while others have defined scanning their surroundings as vigilance (e.g. Cowlishaw 1998; Treves 1999a). Such differences in the definition of vigilance behaviour may be linked to characteristics of the study species, given that the definition of vigilance in primate studies is more difficult than that used in bird studies, which is based on a ‘head-up’ or ‘head-down’ posture. In wild chimpanzees, the clearest definition of vigilance available is that an individual stands up and fixes its gaze on the surrounding environment. However, most cases of such vigilant behaviour have been observed only after a chimpanzee has heard the vocalization of a leopard or of chimpanzees in a neighbouring group; its occurrence was too rare during the focal observations of this study for quantita583

Vigilance Behaviour in Chimpanzees

tive analysis. Therefore, I focused on a ‘lower’ level of vigilance behaviour in this study, and defined vigilance as behaviour that involved gazing fixed on the surrounding environment in a head-up, stationary sitting posture (i.e. it did not include glancing toward food handled by the focal individual, the nearby ground or the individual’s own body, but it did include glances toward group members). When I could not observe the eyes of the focal individual directly, I judged vigilance from the direction of the face or neck of the focal individual. With regard to neighbours, individuals aged <4 yr were not counted. When I observed a female, I also recorded the position of her youngest dependent infant. If the infant was in contact with the mother or within arm’s reach, the situation was recorded as ‘contact’; the situation was recorded as ‘separate’ when the infant was beyond arm’s reach. Activities were classified as foraging, resting, moving, social (e.g. grooming, playing, showing aggression and copulating), and other (e.g. lying in the nest and drinking water). Chimpanzees live in a society characterized by a high degree of fission–fusion and the group’s overall cohesion pattern fluctuates seasonally. At this study site, multiple researchers checked the presence of the M-group individuals in a party in which the focal target individual was attending in each observation day, and the number of individuals in the party was called the ‘daily party size’ (see Mitani & Nishida 1993; Matsumoto-Oda et al. 1998). Dominance Rank

Dominance rank among 13 adult and adolescent males was assessed using ad libitum sampling of the direction of pant-grunt vocalizations (Noe¨ et al. 1980; see Kutsukake 2003). I did not observe pant-grunt vocalizations from any of the adult males toward adult females, which suggested that the adult males were dominant over the adult females. Dominance interactions among females were quite rare (Nishida 1989), which prevented the determination of dominance relationships among them. Consequently, I defined the relatively dominant individuals as follows: for focal males, those males whose dominance rank was higher than that of the focal male; for focal females, all adult males were defined as dominant individuals. The Number of Proximate Individuals and Dyadic Association

During the focal observations, I recorded neighbours every 5 min using the instantaneous sampling 584

N. Kutsukake

method (Altmann 1974) to examine the dyadic association pattern. Consecutive point samples may lack statistical independence when the composition of the group members in proximity does not change for a period that is longer than that of the sampling interval (i.e. 5 mins). To deal with this problem, I determined the minimum sampling interval that was to be treated as statistically independent. I chose two focal males (DG and CT) and two females (AK and OP) randomly and investigated the characteristic bout length using log survivorship curves of association duration (Lehner 1996). I found a break point between the steep and gradual sections of the curve at 5–7 min. Because it is recommended that the time interval selected be slightly longer than that at which the slope changes rapidly, I used data separated by 10-min intervals. Furthermore, for a conservative estimate of the association pattern, I excluded sampling data in which the number of neighbours did not change (Boesch 1991; NewtonFisher 1999). From those data, I calculated the ‘simple ratio index’ in each dyad (Cairns & Schwager 1987). To categorize the association level, I calculated the quartile points of the dyadic association level between unrelated individuals for each focal individual. I defined the dyads whose association score was higher than the third quartile score as ‘affiliative’, those whose association score was less than the first quartile as ‘non-affiliative’, and others as ‘neutral’ (Cords & Aureli 1993). Mother–offspring pairs and sisters were considered affiliative group members. Data Analysis

I excluded the data for periods when I could not observe whether the focal individual was vigilant. For example, I could not collect sufficient data on the vigilance states of focal individuals during moving activity, because I usually followed them from behind and could not see their faces. For the same reason, data were excluded from the analysis when the focal individual was in a nest in a tree. In addition, data after an encounter with a neighbouring group (mainly auditory) or during a patrol or hunting were excluded from the analysis, because those activities apparently influence individual vigilance frequency (e.g. see Baldellou & Henzi 1992; Rose & Fedigan 1995; Gould et al. 1997). Furthermore, I deleted the data when the focal individual was involved in social activity, because social activity includes various behaviours. Therefore, I reported the vigilance behaviour during foraging and resting only. Ethology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

Vigilance Behaviour in Chimpanzees

N. Kutsukake

An adjacent sample taken with the use of the instantaneous sampling method is unlikely to be independent statistically. Nevertheless, another measurement of vigilance behaviour in the study subjects using 2-min focal observations showed that resting chimpanzees are vigilant 2.4 times per min on average (unpubl. data). This suggests that the vigilance interval is far <5 min and one vigilance bout does not exceed 5 min, which justify using vigilance data taken every 5 min as independent data. Furthermore, I used a Generalized Linear Mixed Model (GLMM) to deal with potential pseudoreplication (Schall 1991; Crawley 2002). Mixed models allow both fixed and random components to be fitted to a model. I fitted the identity of the focal individual and the observation day as random terms. Random terms consider repeated sampling within the same focal individuals or the same observation day. GLMMs with binomial error structures and logit link functions were used. This study reports the proportion of data in which a focal chimpanzee was vigilant out of all the data for the focal chimpanzee as the ‘vigilance level’. I analysed data for males and females separately because (1) it is strongly expected that the factors affecting the vigilance level should differ by sex (i.e. an infant influences females only) and (2) when I combined the data for males and females, the best model included a four-way interaction (among the sex of a focal individual, individual activity, the height from the ground and the number of group members in proximity). Such a model is difficult to interpret without separating the data according to one categorical independent term (e.g. sex of the focal individual) and reducing the number of independent terms. The first analyses were designed to investigate the overall occurrence of vigilance behaviour by running GLMM on the entire data sets for males and females separately. Fixed explanatory terms included activity (categorical: foraging or resting), height from the ground (continuous: 0–30 m), number of neighbours within 3 m (continuous: 0–11 individuals), daily party size (continuous: 2–35 individuals) and infant position (categorical: contact or separate, for female analysis only). In accordance with Crawley (2002), I included all likely explanatory variables and possible interactions in the maximal model and excluded terms sequentially until the model only included terms whose elimination would significantly decrease the Akaike Information Criterion of the model. If maternal vigilance varies according to infant position, the predation risk, the conspecific threat or Ethology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

both might have caused the variation (see Introduction). To investigate the factors affecting maternal vigilance, I investigated the influence of infant position on the maternal vigilance level by running separate GLMM analyses on two data subsets, one in which there were no neighbours and the mother– infant pair was on the ground (i.e. high predation risk), and the other in which there was at least one neighbour irrespective of the height from the ground. To test the influence of relationship quality (i.e. the relative dominance relationship or the association level with neighbours) on vigilance level, I chose a data subset for which there was at least one neighbour and analysed males and females separately by simply rerunning the GLMM while fitting the presence or absence of a dominant individual (categorical) and the presence or absence of a non-affiliative group member (categorical) as two additional fixed effects factors. One may argue that relationship quality variables should be set as independent variables in the initial analyses (the entire male and the entire female data sets, analysed separately); however, such analyses give biased estimations of statistical influences for the effects of the number of neighbours on the vigilance level. This is because when there were no neighbours present, the relationship quality variable was automatically excluded from the analyses of the neighbour effect. In other words, the analyses only investigated the influence of the number of neighbours on the vigilance level when there was at least one neighbour. Given that the primary aim of the initial analyses was to investigate how the vigilance level varies with the number of neighbours, including situations in which the focal individual does not have a neighbour, I ran another analysis on a data subset to investigate the influences of relationship quality on vigilance level. In this analysis, I excluded the data for the alphamale, because he lacked a dominant individual. Because the presence of a dominant group member did not influence vigilance level significantly (see Results), I re-ran the statistics after including the data for the alpha-male to increase the sample size and reported that result. Results The GLMM analysis showed that male vigilance was influenced by activity, height from the ground and the number of neighbours but not by daily party size (Table 2). Moreover, the interaction between the height from the ground and the number of neighbours was significant, whereas the other interactions 585

Vigilance Behaviour in Chimpanzees

N. Kutsukake

were not (Table 2). These results indicate that (1) the vigilance level was higher during resting (individual x
SE ¼ 44.5
1.8%) than during foraging (3.7
0.5%), (2) the vigilance level was negatively related to the height from the ground and was highest when males were on the ground (32.4
2.1%), and (3) the vigilance level increased with the number of neighbours, but its slope became less steep when a focal male was on the ground (Fig. 1a). Female vigilance was influenced by activity, height from the ground and infant position but not by daily party size or number of neighbours (GLMM: see Table 2). No interaction was significant. That is, females were vigilant less frequently during foraging (individual x
SE ¼ 4.3
0.4%) than during resting (41.6
1.2%). Vigilance level was highest on the ground (33.6
1.9%) and less when the females were in a tree (Fig. 1b) and was higher when the infant was separated from the mother (23.3
1.2%) than when it was in contact with its mother (13.1
2%; Fig. 2). To test the factors causing the observed variation in the maternal vigilance level according to infant position, I reran the GLMM analyses on two data subsets. The analysis of the data in which the mother–infant pair was on the ground with no group members in proximity indicated that the maternal vigilance level was not affected by infant position (v2 ¼ 0.1, df ¼ 1, p ¼ 0.92). On the other hand, however, the infant position affected the female vigilance level when subset data in which there was at least one neighbour was analysed (Table 3; see below). Table 2: Factors affecting the vigilance level estimated by the instantaneous sampling method Independent term Male Intercept Activity Height from the ground Number of neighbours Height from the ground · the number of neighbours Female Intercept Activity Height from the ground Infant position

Wald (v2) df p

b (SE)

)3.19 2.86 )0.03 0.03 0.005

(0.16) (0.21) 484 (0.015) 5 (0.01) 13.5 (0.001) 11.5

1 1 1 1

<0.0001 0.03 <0.0001 <0.001

)3.26 2.96 )0.19 0.45

(0.29) (0.14) (0.07) (0.13)

1 1 1

<0.0001 0.001 <0.0001

443 10.5 12.2

Generalized Linear Mixed Models (GLMMs) with binomial error structure were used in the analyses. There was significant repeatability of the individual identity and the observation day (both in males and females: p < 0.001).

586

Fig. 1: Influence of height from the ground and number of neighbours within 3 m on vigilance level (individual mean
SE) in males (a) and females (b). Height from the ground and the number of neighbours were classified into each four categories (0–1, 2–5, 6–9, and ‡10 m and no neighbour, and one, two or ‡3 neighbours respectively)

To investigate the influence of neighbour identity (i.e. a dominant or non-affiliative neighbour) on vigilance level, I used the data subsets in which the focal individual had at least one neighbour (see Methods). I found that the presence of non-affiliative neighbours increased the vigilance level in both males and females. In males, vigilance level was influenced by the presence of non-affiliative neighbours in addition to the activity and number of neighbours, whereas the height from the ground became non-significant (Table 3; Fig. 3a). Ethology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

Vigilance Behaviour in Chimpanzees

N. Kutsukake

Fig. 2: Influence of infant position on maternal vigilance level. Vigilance level was calculated from data in which the chimpanzees were on the ground and in the tree. The individual mean
SE is shown

Table 3: Influences of proximate individuals on the vigilance level estimated by the instantaneous sampling method Independent term Male Intercept Activity Number of neighbours Presence of non-affiliative individual(s) Female Intercept Activity Height from the ground Number of neighbours Infant position Presence of non-affiliative individual(s)

b (SE)

)2.37 2.86 0.14 0.37

)2.30 2.84 )0.03 )0.19 0.48 0.31

(0.25) (0.21) (0.06) (0.16)

(0.53) (0.13) (0.01) (0.07) (0.12) (0.15)

Wald (v2)

df

p

203 4.9 5.7

1 1 1

<0.0001 0.03 0.02

465 6.3 8.3 16.1 4.1

1 1 1 1 1

<0.0001 0.01 0.004 <0.0001 0.04

Generalized Linear Mixed Models (GLMMs) with binomial error structure were used in the analyses. There was significant repeatability of the individual identity and the observation day (both in males and females: p < 0.03).

In females, the presence of non-affiliative neighbours increased the vigilance level (Table 3; Fig. 3b). In addition, the activity, height from the ground, the number of neighbours and infant position affected the female vigilance level. This means that the female vigilance level increased in the presence of non-affiliEthology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

Fig. 3: Influence of the presence of non-affiliative group members in proximity and number of proximate individuals on vigilance level (individual mean and 1 SE) in males (a) and females (b). Vigilance level was calculated from data in which the chimpanzees were on the ground and in the tree. ‘Presence’ means the condition in which the non-affiliative group members were in proximity with the focal individual. ‘Absence’ means the condition in which there were no non-affiliative group members in proximity with the focal individual. For comparison, the individual mean with 1 SE of the vigilance level when no group members were in proximity (‘Alone’ on the X-axis) is plotted

ative neighbours, despite the fact that it decreased as the number of neighbours increased (Table 3). The presence of dominant neighbours did not affect the vigilance level in either males or females. Discussion This study found that multiple ecological and conspecific factors influenced the vigilance behaviour of 587

Vigilance Behaviour in Chimpanzees

wild chimpanzees. Chimpanzees were less vigilant during foraging than during resting, which agrees with the results of previous studies in various animals (reviewed in Treves 2000) and suggests that vigilance activity conflicts with other forms of activity in this species. Independent of the influence of activity, the vigilance level of both sexes changed with the height from the ground; vigilance was highest when individuals were on the ground and decreased as the height from the ground increased (Fig. 1a, b).The possibility of a chimpanzee encountering a leopard or being successfully attacked by a leopard is highest when a chimpanzee is on the ground. Thus, this result fits the antipredator hypothesis, which predicts that chimpanzees should be more vigilant when they are on the ground than when they are in trees. Concerning the relationship between the number of neighbours or the daily party size and vigilance, a negative relationship is expected if the chimpanzees are vigilant with respect to predation risk, while a positive relationship is predicted if the vigilance is directed to conspecifics. This study found that the vigilance level during resting and foraging did not decrease with an increase in any of the indexes reflecting the individual in relation to gatherings (number of neighbours or daily party size). Furthermore, male vigilance level increased as the number of group members in proximity and the increase in the vigilance level by the number of neighbours was the lowest when the focal male was on the ground (Fig. 1a). An increase of the vigilance level according to the number of neighbours suggests that the influence of conspecifics is relatively stronger than the influence of predation pressure (Hirsch 2002). However, this result does not necessarily indicate that predation pressure has no influence on the male vigilance level. Although the vigilance level did not decrease with the number of neighbours, a relationship between the height from the ground and the number of proximate individuals may indicate caution with respect to the predation risk to some extent, because males showed a high level of vigilance specifically when they were on the ground with few neighbours (Fig. 1a). The relatively high level of vigilance when males were on the ground with no neighbours in proximity might have resulted from a high level of vigilance for leopards and a low level of vigilance for conspecifics. If so, this result can be interpreted as evidence of an interactive effect of predation risk (Frid 1997). That is, the various independent factors do not independently 588

N. Kutsukake

influence the perception of predation risk; empirically, the influence of one independent factor (e.g. the number of neighbours) is predicted to vary with the influence of other factors (e.g. height from the ground). For example, Frid (1997) showed that female Dall’s sheep were less vigilant when the group size increased, but the decrease in vigilance became less steep near a safe refuge. Therefore, the results of this study indicate that both predation pressure and conspecific factors determine the level of vigilance in male chimpanzees. Finally, the result has shown that the degree of conspecific gathering in the short distance around the individual (i.e. within 3 m) rather than one in the large distance around the individual (i.e. daily party size) influences the vigilance level in chimpanzees. This is similar to the findings in other primate species, although the increases in the number of the neighbours in proximity were negatively related to the vigilance level in other studies (van Schaik & van Noordwijk 1989; Rose & Fedigan 1995; Treves 1998, 1999b, 2000; Steenbeek et al. 1999). These two implications may be due to the weak predation pressure on wild chimpanzees (Goodall 1986; Nishida et al. 2003). The absence of a statistical influence with respect to the number of neighbours on female vigilance can be interpreted in two ways. First, the number of neighbours was not a determinant of female vigilance. Second possibility was that the negative result does not necessarily exclude the conspecific monitoring function, because an increase of the number of neighbours may produce changes in vigilance in the opposite direction, i.e. decreased antipredator vigilance and increased vigilance because of conspecific factors. When these offset each other, vigilance may not vary with the number of neighbours (Treves 1999a; Tchabovsky et al. 2001). Although it is difficult to distinguish these two possibilities, the second possibility may be unlikely as the female vigilance level did not increase with the number of neighbours when the predation risk was low (i.e. chimpanzees were in the tree). Independent of the other factors, maternal vigilance level increased when the infant was separated from the mother (Fig. 2a, b), which suggests an infant protection function for vigilance behaviour (Maestripieri 1993a,b; Steenbeek et al. 1999; Treves et al. 2003; Caro 2005). Although increases in maternal vigilance can be caused by either conspecific threats or predation risk, mother chimpanzees appear to be more concerned about conspecifics than about predation. The maternal vigilance level Ethology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

Vigilance Behaviour in Chimpanzees

N. Kutsukake

increased with infant separation when there was at least one neighbour and the probability of conspecific aggression was high, while vigilance did not increase when there were no neighbours and the probability of predation risk was high. Although few studies have investigated how the identity of proximate individuals influences vigilance, I predicted that the quality of the relationship with an interacting individual influences the vigilance (Watts 1998) and vigilance level increases when a risky individual is nearby. In the current study, vigilance level increased by the presence of non-affiliative neighbours in males and females independent of the influence of other factors (Fig. 3). In males, the vigilance level increased with both the number of neighbours and the presence of non-affiliative neighbours, which again indicates a strong influence of conspecific factors on male vigilance levels. Interestingly, the presence of non-affiliative neighbours increased female vigilance, although the number of neighbours was negatively related to the vigilance level. Then, why do chimpanzees need to increase monitoring when non-affiliative neighbours are nearby? An analysis of the target of vigilance might answer this question directly; unfortunately, it was difficult to estimate the target of monitoring or vigilance in the wild. However, it is highly likely that chimpanzees monitor conspecific behaviour to gather social information or to avoid aggression, given the following. (1) Chimpanzees use various visual gestures (Goodall 1986) and visual cues, such as where other individual are looking during social interactions (Call 2001). (2) It may be important for adult males to survey who is associating with whom. Males that affiliate frequently tend to form coalitions or alliances to attack a rival male, and these coalitions or alliances strongly affect the dominance rank (de Waal 1982; Nishida & Hosaka 1996). Thus, monitoring the social behaviour of a less-associated male, in particular, must be important when these males are nearby because they may be competitors for the dominance position. (3) Chimpanzees may need to monitor conspecific behaviour continuously because aggression occurs suddenly, irrespective of the immediate context in this species (de Waal & Hoekstra 1980). In contrast, this study did not find an influence of relative dominance relationship on vigilance behaviour. The presence of a dominant neighbour did not affect the vigilance level in either males or females. It has been said that the chimpanzee society is characterized by a high level of interindividual tolerance (de Waal 1996), which may weaken the Ethology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

effect of relative dominance relationships on vigilance behaviour (cf. Kutsukake 2003). This study has important implications for studies of vigilance in social animals. As Elgar (1989) noted, it is important to consider various independent factors simultaneously to analyse the vigilance pattern. Unfortunately, previous vigilance studies have tended to underestimate those conspecific factors, which might hinder detailed analysis of conspecific influences on vigilance patterns. In primates and possibly in other animals living in individualized societies in which group membership is relatively stable, great variation is found in the dyadic relationship quality within a group. I expect that this variation in relationship quality affects vigilance behaviour in those species, particularly in species with a high level of within-group competition and risk of conspecific aggression. To further elucidate the effects of conspecifics on vigilance behaviour, future studies should consider the influence of neighbour identity in social animal groups in which individuals are identified and for which detailed records of group members are available. Acknowledgements I would like to thank the Tanzanian Commission for Science and Technology, the Serengeti Wildlife Research Institute, the Mahale Mountains Wildlife Research Centre, and Tanzania National Parks for permitting our research and for their support while I was in Tanzania. I would also like to thank Toshisada Nishida, Kenji Kawanaka, Shigeo Uehara, Kazuhiko Hosaka, Michio Nakamura, Shiho Fujita, James Wakibara, Takahisa Matsusaka and Watongwe research assistants, as well as other colleagues of the research team including Chisa Tokimatsu, for their support in various ways. I would like to thank Toshikazu Hasegawa for their supervision and support over the course of this study, as well as Toshimichi Nemoto and his family for their support. Tim H Clutton-Brock gave critical comments, for which I am grateful. This study was financially supported by Monbusho Scientific Research Fund (Basic Research A1, no. 12375003 to T. Nishida) and JSPS Research Fellowships for Young Scientists. Literature Cited Alberts, S. C. 1994: Vigilance in young baboons: effects of habitat, age, sex, and maternal rank on glance rate. Anim. Behav. 47, 749—755.

589

Vigilance Behaviour in Chimpanzees

Altmann, J. 1974: Observational study of behavior: sampling methods. Behaviour 49, 227—265. Baldellou, M. & Henzi, P. 1992: Vigilance, predator detection and the presence of supernumerary males in vervet monkey troops. Anim. Behav. 43, 451—461. Barbosa, A. 2002: Does vigilance always covary negatively with group size? Effects of foraging strategy. Acta Ethol. 5, 51—55. Beauchamp, G. 2001: Should vigilance always decrease with group size? Behav. Ecol. Sociobiol. 51, 47—52. Beauchamp, G. 2003: Group-size effects on vigilance: a search for mechanisms. Behav. Process. 63, 111—121. Bednekoff, P. A. & Lima, S. L. 1998: Randomness, chaos and confusion in the study of antipredator vigilance. Trends Ecol. Evol. 13, 284—287. Bertram, B. C. R. 1980: Vigilance and group size in ostriches. Anim. Behav. 28, 278—286. Blumstein, D. T., Evans, C. S. & Daniel, J. C. 1999: An experimental study of behavioural group size effects in tammar wallabies, Macropus eugenii. Anim. Behav. 58, 351—360. Boesch, C. 1991: The effects of leopard predation on grouping patterns in forest chimpanzees. Behaviour 117, 220—242. Burger, J. & Gochfeld, M. 1994: Vigilance in African mammals: differences among mothers, other females, and males. Behaviour 131, 153—169. Caine, N. G. & Marra, S. L. 1988: Vigilance and social organization in two species of primates. Anim. Behav. 36, 897—904. Cairns, S. & Schwager, S. J. 1987: A comparison of association indices. Anim. Behav. 35, 1454—1469. Call, J. 2001: Chimpanzee social cognition. Trends Cogn. Sci. 5, 388—393. Cameron, E. Z. & du Toit, J. T. 2005: Social influences on vigilance behaviour in giraffes, Giraffa camelopardalis. Anim. Behav. 69, 1337—1344. Caro, T. 2005: Antipredator Defenses in Birds and Mammals. University of Chicago Press, Chicago. Coolen, I. 2002: Increasing foraging group size increases scrounger use and reduces searching efficiency in nutmeg manikins (Lonchura punctulata). Behav. Ecol. Sociobiol. 52, 232—238. Cords, M. & Aureli, F. 1993: Pattern of reconciliation among juvenile long-tailed macaques. In: Juvenile Primates – Life History, Development, and Behaviour. (Pereira, M. E. & Fairbanks, L. A., eds). Oxford University Press, Oxford. pp. 271—284. Cowlishaw, G. 1998: The role of vigilance in the survival and reproductive strategies of desert baboons. Behaviour 135, 431—452. Crawley, M. J. 2002: Statistical Computing: An Introduction to Data Analysis Using S-Plus. John Wiley, London.

590

N. Kutsukake

Elgar, M. A. 1989: Predator vigilance and group size in mammals and birds. Biol. Rev. 64, 13—33. Frid, A. 1997: Vigilance by female Dall’s sheep: interactions between predation risk factors. Anim. Behav. 53, 799—808. Goodall, J. 1986: The Chimpanzees of Gombe: Patterns of Behaviour. Belknap Press, Cambridge, MA. Goss-Custard, J. D., Cayford, J. T. & Lea, S. G. 1999: Vigilance during food handling by Oystercatchers Haematopus ostralegus reduces the chances of losing prey to kleptoparasites. Ibis 141, 368—376. Gould, L., Fedigan, L. M. & Rose, L. M. 1997: Why be vigilant? The case of the alpha animal. Int. J. Primatol. 18, 401—414. Heathcote, C. F. 1987: Grouping of eastern grey kangaroos in open habitat. Aust. Wildl. Res. 14, 343—348. Hiraiwa-Hasegawa, M., Byrne, R. W., Takasaki, H. & Byrne, J. M. 1986: Aggression toward large carnivores by wild chimpanzees of Mahale Mountains National Park, Tanzania. Folia Primatol. 47, 8—13. Hirsch, B. T. 2002: Social monitoring and vigilance behavior in brown capuchin monkeys (Cebus apella). Behav. Ecol. Sociobiol. 52, 458—464. Hunter, T. B. & Skinner, J. D. 1998: Vigilance behaviour in African ungulates: the role of predation pressure. Behaviour 135, 195—211. Jones, M. 1998: The function of vigilance in sympatric marsupial carnivores: the eastern quoll and the Tasmanian devil. Anim. Behav. 56, 1279—1284. Keverne, E. B., Leonard, R. A., Scruton, D. M. & Young, S. K. 1978: Visual monitoring in social groups of talapoin monkeys (Miopithecus talapoin). Anim. Behav. 26, 933—944. Knight, S. K. & Knight, R. L. 1986: Vigilance patterns of bald eagles feeding in groups. Auk 103, 263—272. Kutsukake, N. 2003: Assessing relationship quality and social anxiety among wild chimpanzees using self-directed behaviour. Behaviour 140, 1153—1171. Kutsukake, N. & Castles, D. L. 2004: Reconciliation and post-conflict third-party affiliation among wild chimpanzees at the Mahale Mountains, Tanzania. Primates 45, 157—165. Kutsukake, N. & Matsusaka, T. 2002: Incident of intense aggression by chimpanzees against an infant from another group in Mahale Mountains National Park, Tanzania. Am. J. Primatol. 58, 175—180. Lehner, P. N. 1996: Handbook of Ethological Methods, 2nd edn. Cambridge University Press, Cambridge. Lima, S. L. & Bednekoff, P. A. 1999a: Temporal variation in danger drives anti-predator behavior: the predation risk allocation hypothesis. Am. Nat. 153, 649—659. Lima, S. L. & Bednekoff, P. A. 1999b: Back to the basics of anti-predatory vigilance: can non-vigilant animals detect attack? Anim. Behav. 58, 537—543. Ethology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

N. Kutsukake

Lima, S. L., Zollner, P. A. & Bednekoff, P. A. 1999: Predation, scramble competition, and the vigilance group size effect in dark-eyed juncos (Junco hyemalis). Behav. Ecol. Sociobiol. 46, 110—116. Maestripieri, D. 1993a: Maternal anxiety in rhesus macaques (Macaca mulatta). I. Measurement of anxiety and identification of anxiety-eliciting situations. Ethology 95, 19—31. Maestripieri, D. 1993b: Maternal anxiety in rhesus macaques (Macaca mulatta). II. Emotional bases of individual differences in mothering style. Ethology 95, 32—42. Matsumoto-Oda, A., Hosaka, K., Huffman, M. A. & Kawanaka, K. 1998: Factors affecting party size in chimpanzees of the Mahale Mountains. Int. J. Primatol. 19, 999—1011. Mitani, J. & Nishida, T. 1993: Contexts and social correlates of long-distance calling by male chimpanzees. Anim. Behav. 45, 735—746. Newton-Fisher, N. E. 1999: Association by male chimpanzees: a social tactic? Behaviour 136, 705—730. Nishida, T. 1989: Social interactions between resident and immigrant female chimpanzees. In: Understanding Chimpanzees (Heltne, P. G. & Marquardt, L. A., eds). Harvard University Press, Cambridge, pp. 68—89. Nishida, T. 1990: The Chimpanzees of the Mahale Mountains – Sexual and Life History Strategies. University of Tokyo Press, Tokyo. Nishida, T. & Hosaka, K. 1996: Coalition strategies among adult male chimpanzees of the Mahale mountains, Tanzania. In: Great Ape Societies (McGrew, W. C., Marchant, L. F. & Nishida, T., eds). Cambridge University Press, Cambridge, pp. 114—134. Nishida, T., Corp, N., Hamai, M., Hasegawa, T., HiraiwaHasegawa, M., Hosaka, K., Hunt, K. D., Itoh, N., Kawanaka, K., Matsumoto-Oda, A., Mitani, J. C., Nakamura, M., Norikoshi, K., Sakamaki, T., Turner, L., Uehara, S. & Zamma, K. 2003: Demography, female life history, and reproductive profiles among the chimpanzees of Mahale. Am. J. Primatol. 59, 99—121. Noe¨, R., de Waal, F. B. M. & van Hooff, J. A. R. A. M. 1980: Types of dominance in a chimpanzee colony. Folia Primatol. 34, 90—110. Quenette, P-Y. 1990: Functions of vigilance in mammals: a review. Acta Oecol. 11, 801—818. Roberts, S. C. 1988: Social influences on vigilance in rabbits. Anim. Behav. 36, 905—913. Robinette, R. L. & Ha, J. C. 2000: Social and ecological factors influencing vigilance by northwestern crows, Corvus caurinus. Anim. Behav. 62, 447—452. Rose, L. M. & Fedigan, L. M. 1995: Vigilance in whitefaced capuchins, Cebus capucinus, in Costa Rica. Anim. Behav. 49, 63—70.

Ethology 112 (2006) 581–591 ª 2006 The Author Journal compilation ª 2006 Blackwell Verlag, Berlin

Vigilance Behaviour in Chimpanzees

van Schaik, C. P. & van Noordwijk, M. A. 1989: The special role of male Cebus monkeys in predation avoidance and its effect on group composition. Behav. Ecol. Sociobiol. 24, 265—276. Schall, R. 1991: Estimation in generalized linear models with random effects. Biometrika 78, 719—727. Smith, A. C., Kelez, S. & Buchanan-Smith, H. M. 2004: Factors affecting vigilance within wild mixed-species troops of saddleback (Saguinus fuscicollis) and moustached tamarins (S. mystax). Behav. Ecol. Sociobiol. 56, 18—25. Steenbeek, R., Piek, R. C., van Buul, M. & van Hooff, J. A. R. A. M. 1999: Vigilance in wild Thomas’s langurs (Presbytis thomasi): the importance of infanticide risk. Behav. Ecol. Sociobiol. 45, 137—150. Tchabovsky, A. V., Popov, S. V. & Krasnov, B. R. 2001: Intra- and interspecific variation in vigilance and foraging of two gerbillid rodents, Rhombomys opimus and Psammomys obesus: the effect of social environment. Anim. Behav. 62, 965—972. Treves, A. 1998: The influence of group size and neighbors on vigilance in two species of arboreal monkeys. Behaviour 135, 453—481. Treves, A. 1999a: Within-group vigilance in red colobus and redtail monkeys. Am. J. Primatol. 48, 113—126. Treves, A. 1999b: Has predation shaped the social systems of arboreal primates? Int. J. Primatol. 20, 35—67. Treves, A. 1999c: Vigilance and spatial cohesion among blue monkeys. Folia Primatol. 70, 291—294. Treves, A. 2000: Theory and method in studies of vigilance and aggregation. Anim. Behav. 60, 711—722. Treves, A., Drescher, A. & Ingrisano, N. 2001: Vigilance and aggregation in black howler monkeys (Alouatta pigra). Behav. Ecol. Sociobiol. 50, 90—95. Treves, A., Drescher, A. & Snowdon, C. T. 2003: Maternal watchfulness in black howler monkeys (Alouatta pigra). Ethology 109, 135—146. de Waal, F. B. M. 1982. Chimpanzee Politics – Power and Sex Among Apes. The Johns Hopkins University Press, London. de Waal, F. B. M. 1996: Conflict as negotiation. In: Great Ape Societies (McGrew, W. C., Marchant, L. F. & Nishida, T., eds). Cambridge University Press, Cambridge, pp. 159—172. de Waal, F. B. M. & Hoekstra, J. A. 1980: Contexts and predictability of aggression in chimpanzees. Anim. Behav. 28, 929—937. Watts, D. P. 1998: A preliminary study of selective visual attention in female mountain gorillas (Gorilla gorilla beringei). Primates 39, 71—78. Zuberbuhler, K. & Jenny, D. 2002: Leopard predation and primate evolution. J. Hum. Evol. 43, 873—886.

591