Social network analysis has been highlighted as a powerful tool to enhance the evidence-based management of captive-housed species through its ability to quantify the social experience of individuals. We apply this technique to explore the social structure and social roles of 50 Livingstone’s fruit bats (Pteropus livingstonii) housed at Jersey Zoo, Channel Islands, through the observation of associative, affiliative, and aggressive interactions over two data collection periods. We implement binomial mixture modelling and characteristic-based assortment quantification to describe the complexity and organisation of social networks, as well as a multiple regression quadratic assignment procedural (MRQAP) test to analyse the relationship between network types. We examine the effects of individual characteristics (i.e., sex, age, and dominance rank) on social role by fitting models to explain the magnitude of node metrics. Additionally, we utilize a quadratic assignment procedural (QAP) test to assess the temporal stability of social roles over two seasons. Our results indicate that P. livingstonii display a non-random network structure. Observed social networks are positively assorted by age, as well as dominance rank. The frequency of association between individuals correlates with a higher frequency of behavioural interactions, both affiliative and aggressive. Individual social roles remain consistent over ten months. We recommend that, to improve welfare and captive breeding success, relationships between individuals of similar ages and dominance levels should be allowed to persist in this group where possible, and separating individuals that interact frequently in an affiliative context should be avoided.

A critical component of conserving and housing species ex situ is an explicit scientific understanding of the physiological underpinnings of their welfare. Cortisol has been repeatedly linked to stress, and therefore used as an indicator of welfare for many species. In order to measure cortisol in the Livingstone's fruit bat (Pteropus livingstonii; a critically endangered keystone species) without disturbing the captive population, we have developed and validated a non-invasive, novel hormone extraction procedure and faecal glucocorticoid assay. A total of 92 faecal samples, 73 from the P. livingstonii breeding colony at Jersey Zoo, Channel Islands and 19 samples from P. livingstonii housed at Bristol Zoological Gardens, UK, have been collected and analyzed. Mixed-effect modelling of the influence of physiological state variables on cortisol concentration revealed that lactating females had higher cortisol levels than non-lactating females, indicating that our assay is measuring biologically relevant hormone concentrations. Males and older bats also had higher cortisol than non-lactating females and younger individuals. Further analysis applied social network methodology to compare the cortisol levels of bats with different social roles. We found that individuals that linked social groups possessed higher than average cortisol levels and conversely, individuals with high-quality, positive relationships had lower cortisol levels. These results demonstrate, for the first time in a bat species, social mediation of stress hormones. Lastly, the frequency of vocalisation was found to positively correlate with cortisol concentration in males, suggesting that this behaviour may be used by animal management as a visual indicator of a bat's hormonal status. Hence, this research has provided unique insights and empirical scientific knowledge regarding the relationship between the physiology and social behaviour of P. livingstonii, therefore allowing for recommendations to be made to optimise bat welfare at the individual level.

Animal behavior and welfare science can form the basis of zoo animal management. However, even basic behavioral data are lacking for the majority of amphibian species, and species-specific research is required to inform management. Our goal was to develop the first ethogram for the critically endangered frog Xenopus longipes through observation of a captive population of 24 frogs. The ethogram was applied to produce a diurnal activity budget and to measure the behavioral impact of a routine health check where frogs were restrained. In the activity budget, frogs spent the vast majority of time swimming, resting in small amounts of time devoted to feeding, foraging, breathing, and (in males) amplexus. Using linear mixed models, we found no effect of time of day or sex on baseline behavior, other than for breathing, which had a greater duration in females. Linear mixed models indicated significant effects of the health check on duration of swimming, resting, foraging, feeding, and breathing behaviors for all frogs. This indicates a welfare trade-off associated with veterinary monitoring and highlights the importance of non-invasive monitoring where possible, as well as providing candidates for behavioral monitoring of acute stress. This investigation has provided the first behavioral data for this species which can be applied to future research regarding husbandry and management practices.