Principal Component Analysis as a Novel Method for the Assessment of the Enclosure Use Patterns of Captive Livingstone’s Fruit Bats (Pteropus livingstonii)

Abstract The Spread of Participation Index (SPI) is a standard tool for assessing the suitability of enclosure design by measuring how captive animals access space. This metric, however, lacks the precision to quantify individual-level space utilization or to determine how the distribution of resources and physical features within an enclosure might influence space use. Here we demonstrate how Principal Component Analysis (PCA) can be employed to address these aims and to therefore facilitate both individual-level welfare assessment and the fine-scale evaluation of enclosure design across a range of captive settings. We illustrate the application of this methodology by investigating enclosure use patterns of the Livingstone’s fruit bat (Pteropus livingstonii) population housed at Jersey Zoo. Focal sampling was used to estimate the time each of 44 individuals in the first data collection period and 50 individuals in the second period spent in each of 42 theoretical enclosure sections. PCA was then applied to reduce the 42 sections to five and seven ecologically relevant “enclosure dimensions” for the first and second data collection periods respectively. Individuals were then assigned to the dimension that most accurately represented their enclosure use patterns based on their highest dimensional eigenvalue. This assigned dimension is hereafter referred to as the individual’s Enclosure Use Style (EUS). Sex was found to be significantly correlated with an individual’s EUS in the second period, whilst age was found to significantly influence individual fidelity to assigned EUS. When assessing the effect of resource location on group-level preference for certain sections, the presence of feeders and proximity to public viewing areas in period one, and feeders and heaters in period two, were positively correlated with space use. Finally, individual EUS remained consistent between both data collection periods. We interpret these results for this species in the context of its observed behavioural ecology in the wild and evaluate the degree to which the current captive enclosure for this population allows for optimal individual welfare through the facilitation of spatial choice. We then explore how these methods could be applied to safeguard captive animal welfare across a range of other scenarios.