Habitat, connectivity, and roadkill of Korea's apex predator, the yellow-throated marten

Context. The yellow-throated marten (Martes flavigula), an endangered species in South Korea, is an important apex predator of that country's temperate forest ecosystem. Human impacts on the landscape continue to threaten marten populations, for which there is insufficient understanding of critical habitats and areas required for population connectivity. Aims. In this study, our aim was to develop models of habitat suitability and connectivity networks for the yellow-throated marten to inform comprehensive landscape conservation strategies. Methods. We used an ensemble of five species distribution models (boosted regression trees, BRT; generalised linear model, GLM; multivariate adaptive regression spline, MARS, maximum entropy model, MaxEnt; and random forest, RF) to identify the main environmental factors influencing marten distribution and potential suitable habitats. We employed a network-based landscape lattice approach combined with circuit theory to predict the potential habitat linkages of martens, and we examined their relevance to marten roadkill. Key results. Five species distribution models performed well with a range of test area under the curve (AUC) values from 0.809 to 0.826 (mean = 0.820) with true skill statistic values over 0.5. Elevation and the amount of vegetation cover were the first and second-most important factors for occurrence probability, and they generally exerted positive influences. Distances from human settlements and roads were also positively related to occurrence probability. We determined appropriate spatial resolutions for conserving functional habitat linkages, which are related to its roadkill distribution, and identified the distribution pattern of key linkage areas. Conclusions. We effectively identified the potential distribution of marten habitats and linkages between them at regional and functional levels to maintain ecological processes. Implications. The results and approaches presented here could be useful in understanding and delineating priority habitats and linkage areas for species conservation and landscape management.