Making better use of tracking data can reveal the spatiotemporal and intraspecific variability of species distributions

Understanding geographic ranges and species distributions is crucial for effective conservation, especially in the light of climate and land use change. However, the spatial, temporal and intraspecific resolution of digital accessible information on species distributions is often limited. Here, we suggest to make better use of high-resolution tracking data to address existing limitations of occurrence records such as spatial biases (e.g. lack of observations in parts of the geographic range), temporal biases (e.g. lack of observations during a certain period of the year), and insufficient information on intraspecific variability (e.g. lack of population- or individual-level variation). Addressing these gaps can improve our knowledge on geographic ranges, intra-annual changes in species distributions, and population-level differences in habitat and space use. We demonstrate this with tracking data and species distribution models (SDMs) of the barnacle goose, a migratory bird species wintering in western Europe and breeding in the Arctic. Our analyses show that tracking data can 1) supplement occurrence records from the Global Biodiversity Information Facility (GBIF) in remote areas such as the European and Russian Arctic, 2) improve information on the temporal use of wintering, staging and breeding areas of migratory species and 3) be used to reveal distribution patterns at the population level. We recommend a broader use of tracking data to address the Wallacean shortfall (i.e. the incomplete knowledge on the geographic distribution of species) and to improve forecasts of biodiversity responses to climate and land use change (e.g. species vulnerability assessments). To avoid common pitfalls, we provide six recommendations for consideration during the research cycle when using tracking data in species distribution modelling, including steps to assess biases and integrate information on intraspecific variability in modelling approaches.