Meta- and subpopulation estimation with disparate data: coconut crabs in the Western Indian Ocean

Widely dispersed fragmented populations are a challenge to monitor because subpopulation sizes may be very small, difficult to access and time consuming to sample regularly. We use the coconut crab (Birgus latro) on Pemba, United Republic of Zanzibar as a case study for estimating highly fragmented populations and metapopulation sizes. The species is a very large terrestrial decapod threatened by exploitation and habitat alteration and now classified as vulnerable. We developed an integrated model to analyse capture-mark-recapture (CMR) data from five sites jointly with count data from 24 sites to estimate site-level densities and population sizes, predicted total population size across the Pemba archipelago, and investigated the effect of six predictors of human influence on density. We fitted separate models to test the effect of the same predictors on raw counts and individual body mass. We estimate the total population of coconut crabs on the Pemba archipelago to be c. 6700 terrestrial individuals. We show that government protection generally affects crabs positively, whereas presence of agriculture negatively affects their densities. This study highlights that time-consuming CMR data can be leveraged to estimate densities on less visited sites, and that fully protected islands are critical for maintaining relatively high population densities. Our overall population estimate suggests that Pemba still hosts a viable coconut crab population in a part of its range where the species is otherwise in steep decline.