Detecting changes in the annual movements of terrestrial migratory species: using the first-passage time to document the spring migration of caribou.

BACKGROUND Migratory species face numerous threats related to human encroachment and climate change. Several migratory populations are declining and individuals are losing their migratory behaviour. To understand how habitat loss or changes in the phenology of natural processes affect migrations, it is crucial to clearly identify the timing and the patterns of migration. We propose an objective method, based on the detection of changes in movement patterns, to identify departure and arrival dates of the migration. We tested the efficiency of our approach using simulated paths before applying it to spring migration of migratory caribou from the Rivière-George and Rivière-aux-Feuilles herds in northern Québec and Labrador. We applied the First-Passage Time analysis (FPT) to locations of 402 females collected between 1986 and 2012 to characterize their movements throughout the year. We then applied a signal segmentation process in order to segment the path of FPT values into homogeneous bouts to discriminate migration from seasonal range use. This segmentation process was used to detect the winter break and the calving ground use because spring migration is defined by the departure from the winter range and the arrival on the calving ground. RESULTS Segmentation of the simulated paths was successful in 96% of the cases, and had a high precision (96.4% of the locations assigned to the appropriate segment). Among the 813 winter breaks and 669 calving ground use expected to be detected on the FPT profiles, and assuming that individuals always reduced movements for each of the two periods, we detected 100% of the expected winter breaks and 89% of the expected calving ground use, and identified 648 complete spring migrations. Failures to segment winter breaks or calving ground use were related to individuals only slowing down or performing less pronounced pauses resulting in low mean FPT. CONCLUSION We show that our approach, which relies only on the analysis of movement patterns, provides a suitable and easy-to-use tool to study species exhibiting variations in their migration patterns and seasonal range use.