National Chung Hsing University, Taiwan (Republic of China)
Abstract: We present a novel fraction community-based site occupancy model to analyze multiple-visit detection/nondetection data, with the goal of disentangling species' mean abundance, occupancy, and detection probability. Our model introduces a fraction community parameter, $c$, which represents the proportion of individuals that are consistently present across visits and ranges from 0 to 1. At $c=0$, the standard occupancy model is recovered, while at $c=1$, the popular N-mixture occupancy model is obtained. We investigate the limitations of these two models through mathematical and simulation studies when the proportion of residents falls between 0% and 100%, which is more likely to occur in practice. Our proposed model provides better data fitting in several occupancy studies, including the fisher (Martes pennanti) species distribution in northern and central California and several bird species from the North American Bird Breeding Survey (BBS).
