Fig. 2From: Mechanistic movement models identify continuously updated autumn migration cues in Arctic caribouThe effect of temperature at three different snow depth levels for each year from the population-level fit of the dynamic-parameter correlated random walk movement model fitted to caribou location data of the Western Arctic Herd, Alaska, 2010–2018. For each year, the predicted effect of temperature (x-axis) on the movement parameter (\(\widehat{{\gamma }_{\text{i}}}\); y-axis) is plotted across 3 levels of snow depth (no snow = 0 cm, average snow = 11 cm, and deep snow = 46 cm). Each black curve is given by the equation \({\widehat{{\gamma }_{\text{i}}}}^{(l)}={logit}^{-1}({{{\varvec{Z}}}_{{\varvec{i}}}}^{{\varvec{T}}}{{\widehat{{\varvec{\beta}}}}_{{\varvec{p}}}}^{\left({\varvec{l}}\right)})\) for the lth Markov-Chain Monte Carlo iteration (termed posterior realizations), and the red line indicates the mean. Annual plots are cut off to the observed range of values for each yearBack to article page