«Item type Thesis or dissertation Authors Davis, Nicolas Citation Davis, N., Schaffner, C. M., & Smith, T. E. (2005). Evidence that zoo visitors ...»
5.4.4 The impact of social events on HPA activity in zoo-housed spider monkeys The present study investigated the impact of social factors on GC levels in a zoo-housed group of spider monkeys using a non-invasive measure of stress: urinary cortisol. Although stress research is used in the assessment of husbandry practices and social relationships in a number of primate species (Honess & Marin, 2006a), there have been only a handful of studies looking at various stressors within a zoo environment (Shepherdson, et al., 2004), and none previously reported in spider monkeys. Earlier research into social stress in primates has predominately examined it from the perspective of dominance hierarchy relationships, which is important in many species and considered a major source of psychological stress (Abbott, et al., 2003; Cavigelli, Dubovick, Levash, Jolly, & Pitts, 2003; Engh, et al., 2006).
However, spider monkeys are reputed to have low or no linear dominance hierarchies (Aureli & Schaffner, 2008) making them an interesting test case for assessing various social factors on their HPA axis responses. In addition, studies of primate aggression have been primarily carried out on species that are largely cohesive, such as macaques, squirrel monkeys, titi monkeys and callitrichids, whereas spider monkey social organisation is characterised by high fission-fusion dynamics (Aureli, et al.,
2008) and therefore the effects of various social factors may have different impacts on their HPA responses. Finally, no previous studies carried out on Ateles have examined the relationship between aggression and their GC response.
The aim of this study was to investigate the impact of social factors on GC levels in a zoo-housed group of spider monkeys using urinary cortisol as a noninvasive measure of stress. The particular categories of social events examined were aggression, reproduction and separation. Based on previous studies, it was believed that these particular events could be associated with an increase in cortisol in zoohoused spider monkeys.
5.5.1 Predictions Based on behaviour studies from the wild (Asensio, et al., 2008; Fedigan & Baxter, 1984; Slater, et al., 2008), and from a survey on aggression in spider monkeys in zoological parks (Chapter 4) I predicted the majority of minor aggression would be carried out by adult males towards receptive adult females when ovulating, or to a lesser extent by adult females to sub adult females as they approach maturity.
Aggression between adult females would not be expected. Any cases of severe or lethal aggression would be expected by adult males towards sub adult males.
Regarding cortisol it would be expected that the largest increases would be associated with the most severe cases of aggression (Ostner, et al., 2008), with severe and lethal aggression associated with the largest increased levels of cortisol. It would also be expected that the response would be dependent on the individuals’ role in the aggressive act with the targets of aggression experiencing the largest increase in urinary cortisol, followed by the actors of aggression and then the bystanders (Creel, 2001; T. E. Smith & French, 1997b). With respect to the timing of cortisol changes, I predicted that cortisol would be elevated the day following aggression and that the rate at which cortisol levels returned to pre-event levels would depend on the severity of the aggression.
Previous studies into reproductive events have shown increases in GCs associated with seasonality and increased competition by males for females (Manson, 2007). However, as spider monkeys are not seasonal breeders (Vick, 2008), and with only one adult male in the study group, competition for females would be at a minimum. Although minor, female-directed male aggression is reported in the wild (Slater et al., 2008; Fedigan & Baxter, 1984), it is unlikely that this would cause a pronounced stress response as it appears to be ritualised in spider monkeys.
However, because the spider monkeys investigated in the present study are zoohoused it may prove difficult for them to engage in species-specific secretive mating (Campbell & Gibson, 2008), therefore I predicted there would be increased GCs in the actors of sexual behaviour at the time of consortships. In addition, postpartum increases in GCs have also been seen in primates in captivity, which have been linked to maternal behaviour (Bahr, et al., 1998; Behringer, et al., 2009) and an increased sensitivity to a stress response from mothers with infants (Boccia, et al., 1995; Maestripieri, et al., 2008).
Severe stress responses have been associated with separations and reintroductions in a number of primate species in captivity (Brent, Kessel, & Barrera, 1997; Clarke, Harrison, & Didier, 1996; Honess & Marin, 2006a; Mendoza, et al., 2000). Therefore, I predicted that separating individuals from the social group would lead to an increase in cortisol levels in the separated individual and to a lesser extent in bystanders. In addition, I predicted that long-term separation would lead to higher cortisol levels than short term separations. Finally, I predicted that reintroduction of group members would also lead to increased cortisol responses for both targets of the separation and bystanders.
5.6.1 Urine samples Urinary cortisol was used as an index of stress because the collection of samples was non invasive, it fitted in well with the daily routine of the group and is a proven method of measuring GCs in primates (Chapter 2). Levels of urinary cortisol were quantified in a total of 2140 samples from six adults present during the study between February 2000 and March 2005 (see Table 5.1).
5.6.2 Event categories For aggression I investigated the intensity, the role of the individual and the timing. Intensity included three levels: minor (aggression which included either no observed injuries or superficial injuries); severe (aggression which included single, or multiple wounds that required veterinary treatment) and lethal (aggression when the individual was killed outright or where the injuries were so serious they necessitated that the individual be euthanized). These levels could be characterised retrospectively following the event and did not rely on the aggression being observed. The role of the individual also included three levels: the target, the actor and bystanders. The role of actor and bystander could only be determined by direct observations of aggressive incidents either by me or by keeping staff which were subsequently recorded. Finally, I also examined the effect of time on aggressive events. This included: prior (samples seven days prior to the event); at (sample from the morning following the event); and post (samples from the following seven days after the event).
For reproduction I investigated the type of event, the role of the individual and the timing of the event. Reproduction included three events: ovulation (signalled on the day by (i) the presence of blood in urine or (ii) the male sniffing in the location of where a female was or had been sitting or (iii) the male handling and sniffing a female’s clitoris); mating (observed copulation between a female and the male) and birth (delivery of an infant following full term pregnancy). The role of the individual in each reproductive event included: the adult male; female target (adult female experiencing the event) and bystanders (other adult females). These events and roles were determined and recorded by myself or by keeping staff throughout the study period as and when they occurred. The sample collection protocol for assessment of reproductive events was identical to that for aggressive events.
For separation I investigated the effect of type of event, the role of the individual and the timing of the event. The type of event included: temporary separation (an individual was out of the group for 24 hr); separation (an individual was out of the group for ≥ 24 hr) and reintroduction (an individual or individuals were reintroduced back into the group following separation). During separations the individual was kept in a section of the enclosure at the back of the exhibit in visual and potential tactile contact with the rest of the group (see Chapter 2, section 2.1.6).
The role of the individual included only two levels: the separated individual and bystanders (individuals not separated from the group). Samples were analysed from the week before to the week after each event. Due to the delay in cortisol being excreted in the urine (see Chapter 1, section 1.4.10) each sample represented the cortisol value for the day prior to the day of collection.
5.6.3 Defrosted samples During storage, a number of samples (763 samples collected between November 2003 and July 2005) were accidentally defrosted over an unknown period of time (as persons unknown unplugged the freezer housing the samples) before they could be assayed. Of these samples, 458 were required for the present study.
Following the discovery of the defrosted event, samples were relabelled to reflect the thaw and then immediately frozen back to -20°C. Cortisol is a cholesterol based steroid hormone and is fairly robust to defrosting (Miki & Sudo, 1998), however due to the unknown time period of the defrosting its effect was investigated. To assess the potential for degradation of the cortisol and check for any potential interference from any other substances that may have formed during the defrosted period, a pool of the defrosted samples (Pool C) was taken and tested against the normal pool (Pool
B) for specificity (see Chapter 2, section 3.2). Pool C was comprised of six samples from each adult and five samples from each sub adult in this social impacts study, comprising a total of 45 samples. This equates to 24 samples from adult females, five from a sub adult female, six from an adult male and 10 from sub adult males. For details on Pool B see Chapter 2, section 3.2. To avoid confounding factors any lower level events which overlapped with any major events were not analysed.
5.6.4 Analyses Factors affecting levels of urinary cortisol were investigated by using linear mixed models (LMM’s). LMM’s allow both fixed and random variables to be fitted to a model, while controlling for variation due to repeated measures of individuals (Tabachnik & Fidell, 2007). The best model was selected by using Akaike’s information criteria (AIC). It compares the adequacy of several models, identifying the model that best explains the variance of the dependent variable as that with the lowest AIC value (Tabachnik & Fidell, 2007). This approach has been used previously in zoo based research on primates were sample size is limited (e.g. O. N.
Fraser, et al., 2008). Maximum likelihood (ML) method was used with fixed variables, and restricted maximum likelihood methods (RELM’s) were used with interactions of fixed variables. An alpha level of 0.05 was adopted for all statistical analysis. The cortisol level was entered as a continuous dependent variable, with identity as the random variable in the models.
The first analysis compared the data samples across types of social event to see whether there were any differences across the event categories (i.e. aggression, reproduction or separation). The subsequent analyses were performed on each type of social event. Post hoc tests used pairwise comparisons using Least Significant Difference. The timing categories for each event were the same as described previously for aggression.
Figure 5.1 % B/Bo of serial dilutions of the Pool B (none defrosted) and Pool C (defrosted) and two cortisol standards to demonstrated parallelism.
pool C at the working dilution of 1:512 and applied to the defrosted samples.
Applying such a conversion value has previously been carried out in a study in which a substance was found to consistently lower the measure of cortisol (Cross, Pines, & Rogers, 2004) I examined the impact of all the stress factors initially in a model to determine whether there were differences in how the different social factors of aggression, reproduction or separation events impacted cortisol values in the spider monkeys. Overall mean values of cortisol for the three categories were 2.408 ± 0.31 SEM, 1.899 ± 0.31 and 1.498 ± 0.35 for aggression, reproduction and separation, respectively. Analyses with LMM’s (ML) (see Table 5.2) revealed that, overall, aggression was responsible for the greatest degree of variance in cortisol levels [F (1, 3) = 10.191, P.0001, AIC = 25940.544] 5.7.1 Aggression A total of 60 aggressive incidents were recorded over the study period. When the actor of aggression could be identified the adult male (Ric) was largely responsible, accounting for 63% of minor cases and 44% of severe cases of Table 5.2 LMM results for the three event categories.
Dependent variables Continuous Cortisol Fixed explanatory variables Event 1 = Aggression, 2 = Reproduction, 3 = Separation Random variables ID
Event [F (1, 3) = 9.92, P.00005, AIC = 22707.01]
aggression. These were carried out predominately towards adult females although incidents of female-female aggression were also recorded, in particular towards the youngest adult female. Finally, there were two cases of lethal aggression which were both carried out by Ric towards juvenile males.
A total of 1709 samples were used in the analysis of 60 incidents of aggression (see Table 5.3). When I examined the impact of different intensities of aggression on the cortisol values for actors, targets, and bystanders the week prior to the day of and the week following the aggressive events a three way interaction of type, role and time provided the best fitting model with the lowest AIC value. [F (1, 23) = 9.772, P.0001, AIC = 9956.406]. The bystander values were calculated using the means of the individual animals so avoiding data pooling (see Table 5.4).
Each type of aggressive event was examined individually to determine where the differences across the three factors of the interaction lay. Cortisol levels for minor incidents (Figure 5.2A) showed very little variation across the role or time, with consistently low levels of cortisol throughout. Of the 29 incidences of minor aggression between adults where the actor was known, the adult male (Ric) was responsible for 20 events that were directed towards adult females. Of the 22 incidences between adults, where the target was known, the youngest adult female (Fay) was the recipient on 14 occasions.