Monitoring the spatiotemporal patterns of wildlife health using rehabilitation databases

Abstract

Wildlife health surveillance is challenging. An alternative is to use wildlife rehabilitation data as potential sentinels, where anomalies in the pattern of submissions may indicate an underlying event that deviates from the baseline and warrants further investigation. Such anomalies may affect multiple species, leading submissions to be clustered in a certain area or time period. To determine spatiotemporal submission patterns and the feasibility of identifying anomalies potentially associated with underlying events, we aggregated databases from two major wildlife rehabilitation centers in Minnesota, drawing on 66,472 submissions from the 2015 – 2019 period, and pertaining to 29 ”species groups” and 12 “circumstances” associated with submission. The infants and juveniles of wildlife, including fledglings, hatchlings, and after-hatch year birds (raptor-specific), submitted as a group from the same location on the same day were grouped and considered as one submission. Hence, the number of records included in the spatiotemporal cluster analysis was 49,352. The multivariate multinomial space-time model of the scan statistic was used to identify statistically significant spatiotemporal clusters of either wildlife species groups or circumstances, simultaneously (Cluster: an area capturing 10% of the submissions aggregated within a maximum radius of 30km during a maximum temporal window of 3-months). This repository contains the data arranged to be used for the spatial cluster analysis.