Two groups of rhesus macaques (group 1, ‘Lewinsky’, and group 2,’Bram’) and two groups of long-tailed macaques (group 3, ‘Lixa’, and group 4, ‘Tofa’) were studied. Groups 1 to 4 included 10, 10, 15 and 16 adult females, respectively, and 7, 4, 7 and 5 adult males. Only sexually mature monkeys, i.e. females older than 3.5 years and males older than 3 years, were included in the analysis. All males were intact, and the age of the non-alpha male subjects ranged between 3 and 26 years (N = 19, mean = 7.7, SD = 5.7) comprising the prime time of macaque males between their 6th and 11th years (Bercovitch et al. 2003). All individuals had several potential non-kin (i.e. from another matriline) sexual partners in the group. Each subject contributed to the dataset as a potential part of a sexual dyad and as a possible bystander individual. Within each group, the hierarchy was determined based on submissive behaviours using Matman 1.1 (de Vries et al. 1993). In all four groups, the dominance hierarchy was significantly linear [groups 1 to 4, respectively: h’ = 0.53, 0.50, 0.55, 0.82, p = 0.0006, 0.0001, 0.0004, 0.0001 (de Vries 1995)]. Rhesus macaques are seasonal breeders, and their mating season in our colony lasts from October to March. Long-tailed macaques may be seasonal breeders (van Schaik and van Noordwijk 1985), yet in our colony are non-seasonal and mate all year round.
The rhesus and long-tailed macaques were housed in similar cages. The cages consisted of interconnected indoor and outdoor enclosures. Enclosures were provided with permanent enrichment consisting of fire hoses, tires, ladders, tree trunks and a swimming pool (Vernes and Louwerse 2010). Monkeys were only fed in the indoor enclosure, outside observation hours. However, water was available ad libitum in the outdoor enclosure.
During the observations, only the outdoor enclosures were available to the monkeys. The outdoor enclosures measured for groups 1 to 4: 260, 270, 218 and 502 m2, respectively, and were of approximate triangular shape (Fig. 1 in Overduin - de Vries et al. 2012, 2013). Compared to groups 1, 2 and 3, group 4 had access to a double-sized outside enclosure (i.e. group 4 used two connected outdoor enclosures).
The groups were observed by five different observers (because of its larger outdoor area, group 4 was observed by two observers simultaneously). The groups were observed over different periods: group 1, 25-11-2008 to 17-02-2009; group 2, 25-11-2008 to 24-02-2009; group 3, 02-02-2010 to 01-04-2010; and group 4, 18-02-2011 to 15-04-2011. The inter-observer reliability of the general scan-taking was checked during a 2-h simultaneous observation of group 4 by two observers. To this end, general proximity scans were made every 5 min. When the exact poles for all subjects in 24 scans of the two observers were compared, the agreement was marginal (kappa = 0.64, index of concordance = 0.66). However, if scoring two adjacent poles is counted as an agreement (which corrects for situations where an individual is sitting equidistantly between two poles, or moves from one pole to another during scoring), the agreement was amply sufficient (index of concordance = 0.82). Sexual behaviours were recorded ad libitum. For long-tailed macaques, which have only single-mount copulations, copulations were scored if they involved three or more thrusts. For rhesus macaques, which have multiple-mount copulations, only the first mount within a mount series was used for analysis. The remainder of the sexual behaviours (sexual present, taillift, waistgrasp and mount) were, for both species, in conformity with the ethogram of our previous studies (Overduin - de Vries et al. 2012, 2013). One or two observation sessions of typically 2 h each, and only occasionally 1 h (mean = 2.0, SD = 0.06), were conducted per day, resulting in respectively 160, 192, 142 and 73 h of ad libitum sampling for the four study groups.
Screens providing hiding locations
We provided screens on a temporary basis and at different locations, so the animals could not use operant conditioning to copulate at a particular spot but had to be flexible when using the screens. The screens were provided to the monkeys just before the start of the first observation session of a given day. Each screen remained at a particular location for 24 h, except when mentioned otherwise. All groups had access to two screens at the same time, precluding the simultaneous monitoring of both sides of both screens by a single bystander monkey, while the screens were oriented along the line of sight of the human observer. The two screens were semi-randomly alternated between six possible locations.
Within the rhesus macaque groups, four different types of screens were created: (a) fully opaque, (b) transparent bottom, opaque top, (c) opaque bottom, transparent top, and (d) fully transparent. The dimensions of the screens (width × height) were 1.2 × 1.3 m. During each day, two different screens were simultaneously available to the monkeys. These screens were used during 42 observation days in group 1 and for 21 observation days in group 2. The screens in group 2 were replaced by two equal, fully opaque wooden screens measuring 2.8 × 1.13 m for further 31 observation days. These larger screens in group 2 were not removed after 24 h but remained in the group permanently.
Since the rhesus macaques hardly ever used the screens, we simplified the set-up of the experiment for the long-tailed macaque study and used only the screen type we presumed to be most effective. In group 3, two equal, fully opaque screens, measuring 3.0 × 0.8 m, were hung from the ceiling. Each of the screens was equipped with a sitting beam on each side of the screen. In addition, several equally elevated beams were provided without screens. This way, a possible preference for the screens would not be caused by a preference for elevated beams.
Since the fully opaque screens in group 3 were hardly used, we adjusted the set-up for the possibility that the macaques required a larger enclosure to show hiding behaviour and/or preferred another screen type that allowed monitoring of the audience while hiding. For group 4, one of the screens was similar to those used in group 3, but we added one peek-hole screen. This peek-hole screen measured 3.0 × 0.8 m and had 51 holes of 5-cm diameter. Both screen types were simultaneously available to the monkeys.
To calculate inter-individual distances, we took scans under two different conditions: (1) ‘baseline behaviour’, in which the location of all animals during non-sexual behaviour is measured, and (2) ‘sexual behaviour’, in which the location of the animals involved in sexual behaviour, and the location of the alpha male are measured. The effect of the alpha male is expected to be strongest, since reported disruptions almost exclusively involve higher ranking individuals affecting lower ranking ones (Chapais 1983; Manson 1996). Moreover, in previous studies, we found that the alpha male has the strongest audience effect on its group members’ sexual behaviour (Overduin - de Vries et al. 2012, 2013). Therefore, the method used in the rhesus macaque study only determined the location of the alpha male in relation to the copulating couple. However, since we found audience effects of other individuals in our previous study, we decided to adjust the method in the subsequent long-tailed macaque observations, so that the location of all sexually mature group members could be determined under both conditions.
Baseline scans were taken at the start and end of each observation session for rhesus macaques, and every half hour for long-tailed macaques. All rhesus macaque scans consisted of individual locations indicated on a scaled map. All long-tailed macaque scans were taken by noting down for each individual the nearest pole of the enclosure. Each cage contained 0.22 vertical poles/m2. In addition, a monkey was noted as hiding near a screen if it was a rhesus macaque within 1 m from the screen or if it was a long-tailed macaque positioned on one of the beams attached to the elevated screen.
The strength of the effect of a particular bystander on the positioning of its group members’ sexual behaviour was calculated by taking the mean difference between baseline and sexual behaviour in distance from a particular bystander. This mean difference was taken as the dependent variable in a general linear model (GLM), and corrected for repeated measures from the same individuals, by incorporating male and female identity as random factors in the model. The estimated marginal mean resulting from the general linear model is hereinafter termed the ‘audience effect strength’.
For long-tailed macaques, we matched group cohesion between sexual and baseline behaviour (this was impossible for rhesus macaques, because we did not have data on group cohesion during sexual behaviour for this species). For each sexual behaviour scan, the mean inter-individual distance (MID) of the individuals that were not involved in sexual behaviour was calculated. For each sexual scan, a baseline scan was selected which had the nearest MID value. Subsequently, the distance of the copulating couple in the sexual scan and the selected baseline scan were compared. A single baseline scan was included in the analysis one to four times, but baseline scans were only repetitively included if both individuals involved in sexual behaviour in the matched sexual scans differed. Hence, each comparison of distance between sexual and baseline scan is unique.
Most tests were conducted using the software package ‘R’ version 2.10. The GLM analyses were executed using SPSS version 20. All statistical tests were two-tailed with α set at 0.05.