Hierarchical state-space models improve estimation of behavioural states from animal movement data

Hierarchical state-space models offer a direct and powerful way to scale inferences of animal movement behaviour from individuals up to populations. Despite this, implementations of hierarchical state-space models are rare, but for highly error-prone data may be the only option for reliable inferences of animal movement and behaviour. I use a combination of simulated and real animal movement paths to assess the relative abilities of hierarchical and non-hierarchical state-space models to infer behavioural states associated with different movement patterns. Movement paths were simulated such that location errors mimicked either GPS or Argos satellite data. Behavioural state estimation error was strongly affected by the degree of similarity between movement patterns characterising the behavioural states, with less error when movements were strongly dissimilar between states. A hierarchical state-space model generally improved behavioural state estimation relative to a non-hierarchical version of the same model and this improvement was greatest for simulated data with heavy-tailed Argos location errors. When applied to Argos telemetry datasets from 10 Weddell seals, the non-hierarchical model estimated highly uncertain behavioural state switching probabilities for most individuals whereas the hierarchical model's estimates had substantially less uncertainty. As a consequence, the hierarchical model was better able to resolve the behavioural state sequences across all seals. I conclude that hierarchical state-space models should be the preferred choice for estimating behavioural states from animal movement data, especially when location data are error-prone.