«Item type Thesis or dissertation Authors Davis, Nicolas Citation Davis, N., Schaffner, C. M., & Smith, T. E. (2005). Evidence that zoo visitors ...»
4.2.2 Analyses To ascertain whether the distribution of the age/sex class of the actors or the targets of aggression differed from a chance distribution chi-square tests were used for goodness-of-fit. The expected frequencies of aggression were weighted by correcting the expected values by the proportion of individuals in each age/sex class in the population to reflect the opportunity for aggression. To minimise the potential for violating the underlying assumption of independence for chi square tests, individual spider monkeys were counted only once for each category of aggression.
This was a conservative approach as it had the effect of underestimating the incidences of aggression by individuals who show repetitive aggressive behaviour and so restricting the sample size. When sample sizes were less than six per cell the Yates correction was applied (Schwiegert, 1994). To assess whether there was a relationship between the density of animals in the enclosure and aggression, the densities of 13 enclosures were calculated based on the modal number of animals in the group over the course of the study and the area of the combined indoor and outdoor enclosure. Pearson’s correlations were used to test for the relationship between density and mean number of aggressive events. The conventional alpha level of 0.05 was adopted for all tests.
Thirty two of the 55 questionnaires were returned (58%), although data from eight zoological parks (14.5%) had to be omitted because the required group composition had changed and no longer met the minimum requirements for my study, or insufficient information was provided about the overall group composition.
Therefore, information regarding 26 different groups of spider monkeys from 24 zoological parks was provided. There were seven groups (26.9%) in which no aggression was reported, which were subsequently referred to as non-aggressive groups, and 19 (73.1%) groups in which aggression was reported, which were subsequently referred to as aggressive groups. The modal overall composition for the data set (n = 26) was one adult male and two adult females. The modal composition for non-aggressive groups (n = 7) was one adult male and one adult female, while the modal composition for aggressive groups (n = 19) was one adult male, two adult females and one non-adult male. Infants were not included in the analyses as they are not involved in any aggressive interactions.
Four species were represented in the survey data A. belzebuth (n = 7), A.
paniscus (n = 3), A. geoffroyi (n = 6) and A. fusciceps (n = 10). The species A.
fusciceps were kept in the largest overall group sizes (including infants) and A.
geoffroyi were ascribed the highest number of aggressive incidents (Table 4.1).
However, chi-square tests did not identify a significant difference in the number of aggressive events among the different species for the actors (χ²(3) = 1.98; P 0.05) or targets of aggression (χ²(3) = 4.07; P0.05), therefore the entire dataset, regardless of species, was pooled and analysed together.
Although information from 143 incidents was obtained, for the investigation into actors and targets, only data where the identities of either the targeted individuals (targets) or the actors of aggression (actors) were known were analysed.
This yielded 56 events for the actors and 127 events for the targets of aggression.
4.3.1 Frequency of aggression Aggression was observed in the majority of spider monkey groups (see Table 4.2), with tension between males identified as the most common context of aggression. The majority of the respondents considered the nature of spider monkey aggression to be different to that of other primates housed within their zoological park. Specifically, they reported aggressive bouts were less frequent, usually occurred without any obvious signs of previous tension and often resulted in more Table 4.1 Composition of groups by species/subspecies for overall modal group size, modal number of adults and non-adults, modal number of males and females and total number of aggressive incidents.
severe injuries to the target animals. They also reported that aggression often occurred when younger males approached breeding age.
Using the criteria described above, 127 cases of aggression were recorded, with 93 cases (73.2%) of minor aggression, 28 cases (22.0%) of severe aggression and six cases (4.7%) of lethal aggression reported.
4.3.2 Actors and intensity of aggression Overall, I found that males were much more likely to be aggressors than females (χ² (1) = 26.18; P 0.001). When the data were examined further by age class, it was found that adult males were the most frequent actors representing 37 cases (66.1%), followed by sub-adult males who accounted for 11 cases (19.6%) and adult females who accounted for 8 cases (14.3%). There were no instances of nonadult females or juvenile males as actors of aggression (see Table 4.3 and 4.4).
Analysis of all the intensities of aggression revealed an overall significant difference for actors (χ² (5) = 32.31; P 0.001). The finding was driven by the adult males who Table 4.2 Summary of response to the questionnaire showing number of responses for each question and proportion of answers.
were more likely to be aggressors than juvenile males (χ² (1) = 12.41; P 0.001), adult females (χ²(1) = 19.09; P 0.001), sub-adult females (χ² (1) = 13.16; P 0.001) and juvenile females (χ² (1) = 14.48; P 0.001), and by sub-adult males who were more likely to be aggressors than juvenile males (χ² (1) = 10.35; P 0.001), adult females (χ²(1) = 17.03; P 0.001), sub-adult females (χ² (1) = 11.10; P 0.001) and juvenile females (χ² (1) = 12.42; P 0.001). There was an overall difference in the distribution for actors of minor aggression from the expected distribution across the six age/sex class categories (χ² (5) = 19.21; P 0.005). Age/sex pairwise comparisons yielded significant differences from the expected distribution as subadult males were more likely to be the actors of mild aggression than adult males (χ² (1) = 11.35; P 0.001), juvenile males (χ² (1) = 8.27; P 0.005), adult females (χ² (1) = 10.88; P 0.001), sub-adult females (χ² (1) = 9.35; P 0.005) and juvenile females (χ² (1) = 9.60; P 0.005). When severe aggression was examined there was also Table 4.3 Summary of reported aggressive incidents towards adult male, sub-adult male and juvenile male spider monkeys as classified by age, sex and intensity.
more of a difference in the distribution of actors across the six age/sex classes than expected (χ² (5) =15.20; P 0.01). The significant difference from the expected distribution was driven by the adult males as they were more likely than expected to target severe aggression compared to sub- adult males (χ² (1) = 6.38; P 0.025), juvenile males (χ² (1) = 7.58; P 0.01), adult females (χ² (1) = 7.16; P 0.01), subadult females (χ² (1) = 8; P 0.005) and juvenile males (χ² (1) = 7.42; P 0.01).
Finally, the distribution for actors of lethal aggression differed from an expected distribution across the four age/sex classes (χ² (5) = 19.12; P 0.001). The difference was due to adult males being more likely than expected to be responsible for lethal aggression compared to sub-adult males (χ² (1) = 7.26; P 0.01), juvenile males (χ² (1) = 7.68; P 0.01),adult females (χ² (1) = 8.48; P 0.005) sub-adult females (χ² (1) = 8.38; P 0.005) and juvenile females (χ² (1) = 7.36; P 0.01).
4.3.3 Targets and intensity of aggression Overall, males were more likely than expected to be the targets of aggression, regardless of intensity (χ² (1) = 18.29; P 0.001). Adult females were the targets of aggression in 46 cases (36.2%) of the aggressive bouts reported. Adult males were the targets of aggression in 25 cases (19.7%), juvenile males in 20 of the cases (15.7%), non-adult females in 14 of the cases (11.0%) and sub-adult males were targets of aggression in 12 of the cases (9.4%) (see Table 4.3 and 4.4). The distribution across age/ sex categories for all aggression differed from the expected distribution (χ² (3) = 81.19; P 0.001), however the significant effect was attributable to juvenile males who were more likely to be targeted than expected compared to adult males (χ² (1) = 81.27; P 0.001), sub-adult males (χ² (1) = 84.50;
P 0.001), adult females (χ² (1) = 90.31; P 0.001), sub-adult females (χ² (1) = 90.13; P 0.001) and sub-adult females (χ² (1) = 81.60; P 0.001).
When the distribution, according to the different intensities, was examined more variation in the pattern was identified. The distribution across age/sex categories for minor aggression differed from the expected distribution (χ² (5) = 49.70; P 0.001). The deviation from the expected distribution was due to juvenile males receiving more mild aggression than expected compared to adult males (χ² (1) = 28.80; P 0.001), sub-adult males (χ² (1) = 31.86; P 0.001), adult females (χ² (1) = 31.84; P 0.001), sub-adult females (χ² (1) = 39.69; P 0.001) and juvenile females (χ² (1) = 29.56; P 0.001). The distribution for targets of severe aggression also differed from the expected distribution across the four age/sex classes of individuals (χ² (5) = 32.24; P 0.001). The deviation also differed due to juvenile males receiving more severe aggression than expected compared to adult males (χ² (1) = 37.00; P 0.001), sub-adult males (χ² (1) = 36.76; P 0.001), adult females (χ² (1) = 39.73; P 0.001), sub-adult females (χ² (1) = 36.75; P 0.001) and juvenile females (χ² (1) = 39.00; P 0.001). Finally, the distribution for targets of lethal aggression also differed from expected (χ² (5) = 40.99; P 0.001). Juvenile males were more likely than expected to be targets of lethal aggression than adult males (χ² (1) = 30.58; P 0.001), sub-adult males (χ² (1) = 31.29; P 0.001), adult females (χ² (1) = 34.86; P 0.001), sub-adult females (χ² (1) = 32.92; P 0.001) and juvenile females (χ² (1) = 32.33; P 0.001).
4.3.4 Animal density No correlation between the density of spider monkeys and total mean number of aggressive events (r = 0.08; N = 127; p = 0.81) was found. Furthermore, no significant correlations were found between density and mean number of minor aggressive events (r = 0.08; N = 93; p = 0.79), severe aggressive events (r = 0.03; N = 28; p = 0.93) and lethal aggressive events (r = -0.13; N = 6; p = 0.67).
4.3.5 Context of aggression Of the 143 cases of aggression a detailed explanation was provided by the respondents for 54 (37.8%) of the incidents, and these were separated into five distinct contexts. The most frequent context for aggression was tension among males within the same group, defined as a long-standing situation between two males in which they had repeated conflicts that were not resolved, and was reported by 10 zoological parks and accounted for 50% of the incidents described. The context of introduction, when animals were reintroduced to the group after a period of separation or when new animals were introduced to the group, was reported by seven zoological parks accounting for 25.9% of the incidents. ‘Bullying’ when the whole group chased and harassed an individual, was reported in four zoological parks accounting for 20.4% of the incidents described by zoo keepers. The least frequent context of aggression was public feeding reported by one zoological park accounting for 3.7% of the incidents. General aggression in the group, aggression by males and aggression by females were also reported, but not enough detail was provided to indicate a specific context.
The use of questionnaires as a tool to quantify animal behaviour has been widely used in both applied and theoretical contexts but rely on the fundamental assumptions that the person caring for the animal has access to valid information about the animal’s typical behaviour, and that this information can be extracted in a form that is reasonably reliable and accurate (Gosling, Kwan, & John, 2003; Hsu & Serpell, 2003). The data collected in this study were based on animal records that zoo staff members make on a daily basis, rather than notes from diary entries or memory which were deemed less reliable. To further improve the quality of the data only zoos that were accredited to ISIS, which are committed to keep records to a high standard, were used. Although no pilot study was carried out prior to the questionnaire being distributed its design was based on a previous study into aggression in zoo housed lion tamarins (Inglett, et al., 1989). However, a prototype may have been useful in refining and modifying the questions which may have assisted in the interpretation of the data provided.
The questionnaire provided information about 143 aggressive incidents which were used to evaluate the predictions. Four species of spider monkey were represented in the questionnaire data and although there was variation in the total numbers of incidences of aggression across the different species analyses revealed no significant difference. This allowed the dataset to be pooled. These differences are likely to be attributable to differences in demographics and group size, and that most of the aggressive incidents occurred in a small proportion of the groups. The first prediction was that adult males would be the most frequent actors of minor aggression and that adult females would be the most frequent targets of minor aggression. Although adult males were responsible for the majority of the reported minor aggression, with 25 of the 40 individual cases, and adult females received 41 of the 93 cases of minor aggression, when the proportion of age and sex classes were considered as part of the overall population the results did not support either of these predictions. It was the sub-adult males who were more frequently the actors and juvenile males more frequently the targets of minor aggression than expected. This differs from reports in the wild, which indicate female-directed male aggression is the most frequently reported aggression (Campbell, 2003; Fedigan & Baxter, 1984;
Slater, et al., 2008). Reports of minor aggression between males are virtually absent from wild populations. For example, van Roosmalen and Klein’s (1988) review does not include any reports of male-male aggression.