A Spatiotemporal Bayesian Model for Population Analysis

Abstract

Spatiotemporal population analysis based on incomplete, redundant, and unidentified observations is critically important, yet it is a very challenging problem. Different approaches have been proposed and several methods have been implemented to address this problem. Capture-recapture methods have been widely used and have become the standard sampling and analytical framework for ecological statistics with applications to population analysis. Despite the fact that capture-recapture methods have been commonly used, these methods do not consider the spatial structure of the population. Moreover, conventional capturerecapture methods do not use any explicit spatial information with regard to the spatial nature of the sampling and spatial distribution of individual encounters. Recently a spatial capture-recapture method has been introduced by Royle and Chandler to link observed encounter histories of individuals to spatial population ecology and study the population using new technologies such as remote cameras and acoustic sampling. The first objective of this study was investigating feral hog population in the Kennedy Space Center (KSC) which is part of Merritt Island National Wildlife Refuge in Titusville, about 60 miles east of city of Orlando in Brevard and Volusia Counties in Florida. Due to the size of KSC and the limited resources, two study sites within KSC were chosen for investigation, monitoring, and data collection. These sites were: 1. Happy Creek (HC); and 2. Tel-4. We estimated the hog population using the spatial capture model introduced by Royle and Chandler. The estimated hog population for HC was between 55 and 108 hogs. The estimated hog population for Tel-4 was between 61 and 114 hogs. To estimate the hog population in KSC, we combined the results obtained from two study sites within KSC. We calculated and assigned specific weights to the estimated hog populations in HC and Tel-4 based on their percentage areas in comparison with the entire area of KSC. As a result, the calculated weights were 0.73 and 0.27 for HC and Tel-4 respectively. The estimated hog population N using the proposed weighted averaging was between 3, 058 to 5, 862 hogs. Although the spatial capture method is promising, the estimated population size is not robust and suffers from spatial complexity. Therefore, the second objective of this research was to perform a comprehensive study of the parameters of the spatial capture model and their impacts on the estimated population size. The third goal was focused on identification of parameters with significant impact on the estimated population size and to develop informative priors for the identified parameters. The fourth objective was to improve the spatial capture model by integrating camera spatial locations and regularizing spatiotemporal parameters for the estimation of the population size.