Evaluating habitat connectivity methodologies: a case study with endangered African wild dogs in South Africa

Abstract

In fragmented landscapes, connectivity between subpopulations is vital for species’ persistence. Various techniques are used to assess the degree of connectivity between habitat patches, yet their performance is seldom evaluated. Models are regularly based on habitat selection by individuals in resident populations, yet dispersers may not require habitat which supports permanent residence. Using a database of African wild dog (Lycaon pictus) occurrence records in north-eastern South Africa (n = 576), we developed and compared ecological niche models (ENM) for wild dogs packs and dispersers. Additionally, we used least cost path (LCP) and current flow models to assess connectivity. Results were further validated using occurrence records (n = 339) for cheetah (Acinonyx jubatus). The ENM for wild dog packs identified large but isolated patches of suitable habitat, while the disperser ENM had greater suitability values for areas in between highly suitable patches. Without disperser-specific data, models omitted large areas which were confirmed to have provided connectivity. Although models derived from a potentially subjective cost layer have been criticised, the current flow model outperformed the other connectivity techniques and provided the most meaningful predictions for conservation planning. We identified five priority conservation areas for wild dogs, two of which had a greater feasibility for recolonisation. The scarcity of disperser-specific data promotes models using habitat data for resident individuals but here we illustrate the pitfalls thereof. Our study provides insights into the performance of these frequently employed techniques and how they may affect conservation management decisions.