Abstract
Background
Habitat loss is a key threat to the survival of many species. Habitat selection studies provide key information for conservation initiatives by identifying important habitat and anthropogenic characteristics influencing the distribution of threatened species in changing landscapes. However, assumptions about the homogeneity of individual choices on habitat, regardless of life stage, are likely to result in inaccurate assessment of conservation priorities. This study addresses a knowledge gap in how animals at different life stages diverge in how they select habitat and anthropogenic features, using a free-ranging population of African wild dogs living in a human-dominated landscape in Kenya as a case study. Using GPS collar data to develop resource selection function and step selection function models, this study investigated differences between second order (selection of home range across a landscape) and third order (selection of habitat within the home range) habitat selection across four life history stages when resource requirements may vary: resident-non-denning, resident-heavily-pregnant, resident-denning and dispersing.
Results
Wild dogs showed strong second order selection for areas with low human population densities and areas close to rivers and roads. More rugged areas were also generally selected, as were areas with lower percentage tree cover. The strength of selection for habitat variables varied significantly between life stages; for example, dispersal groups were more tolerant of higher human population densities, whereas denning and pregnant packs were least tolerant of such areas.
Conclusions
Habitat selection patterns varied between individuals at different life stages and at different orders of selection. These analyses showed that denning packs and dispersal groups, the two pivotal life stages which drive wild dog population dynamics, exhibited different habitat selection to resident-non-breeding packs. Dispersal groups were relatively tolerant of higher human population densities whereas denning packs preferred rugged, remote areas. Evaluating different orders of selection was important as the above trends may not be detectable at all levels of selection for all habitat characteristics. Our analyses demonstrate that when life stage information is included in analyses across different orders of selection, it improves our understanding of how animals use their landscapes, thus providing important insights to aid conservation planning.
Order Selection of Spatial and Temporal Autoregressive Models with Errors in Variables