Fig. 2

Estimated a posteriori hidden states (using the Viterbi algorithm) of B05-2020 based on a hidden Markov model fit to three response variables: step length, residence time, and day home range. The red hidden state corresponds to the ‘non-movement’ latent state associated with parturition, the grey state corresponds to the ‘low-movement’ latent state, and the white state corresponds to the ‘high-movement’ latent state. Day home range has been truncated at 100 Ha to improve scale. Using our decision rule, we identified a lambing date of May 20, 2020 (red line) which matched the lambing date estimated from field observations