Flexible hidden Markov models for behaviour-dependent habitat selection

Abstract

There is strong incentive to model behaviour-dependent habitat selection, as this can help delineate critical habitats for important life processes and reduce bias in model parameters. For this purpose, a two-stage modelling approach is often taken: (i) classify behaviours with a hidden Markov model (HMM), and (ii) fit a step selection function (SSF) to each subset of data. However, this approach does not properly account for the uncertainty in behavioural classification, nor does it allow states to depend on habitat selection. An alternative approach is to estimate both state switching and habitat selection in a single, integrated model called an HMM-SSF. We build on this recent methodological work to make the HMM-SSF approach more efficient and general. We focus on writing the model as an HMM where the observation process is defined by an SSF, such that well-known inferential methods for HMMs can be used directly for parameter estimation and state classification. We extend the model to include covariates on the HMM transition probabilities, allowing for inferences into the temporal and individual-specific drivers of state switching. We demonstrate the method through an illustrative example of African zebra (Equus quagga), including state estimation, and simulations to estimate a utilisation distribution. In the zebra analysis, we identified two behavioural states, with clearly distinct patterns of movement and habitat selection ("encamped'' and "exploratory''). In particular, although the zebra tended to prefer areas higher in grassland across both behavioural states, this selection was much stronger in the fast, directed exploratory state. We also found a clear diel cycle in behaviour, which indicated that zebras were more likely to be exploring in the morning and encamped in the evening. This method can be used to analyse behaviour-specific habitat selection in a wide range of species and systems. A large suite of statistical extensions and tools developed for HMMs and SSFs can be applied directly to this integrated model, making it a very versatile framework to jointly learn about animal behaviour, habitat selection, and space use.

Time-varying habitat selection analysis: a model and applications for studying diel, seasonal and post-release changes

“…The use of segmentation-clustering algorithm such as segclust2d or hidden-Markov models on the time-series coefficients obtained by time-varying HSA will allow one to do this, and to estimate the duration and frequencies of such modes. This could complement very recent works developing behavioral-mode detection approaches based on step-selection functions (Prima et al 2022; Klappstein, Thomas, and Michelot 2022).…”

Evaluating goodness-of-fit of animal movement models using lineups

“…These states allow for a mixture of movement types (long and directed vs. short and tortuous), which in turn induces correlation between step lengths and turn angles (Hodel and Fieberg 2022). Building on these two approaches, one might next consider fitting a state-switching step-selection function (Klappstein et al 2023, Pohle et al 2023), though the software for fitting and simulating from this class of models is less well developed.…”

“…Integrated step-selection analyses (ISSAs) (Avgar et al 2016, Fieberg et al 2021) and hidden Markov models (HMMs) (Langrock et al 2012, McClintock et al 2012), in particular, are extremely popular for analyzing wildlife telemetry data due to the availability of open-source software () for implementing them (McClintock and Michelot 2018, Signer et al 2019). These approaches, as well as recently developed methods that make it possible to combine the two frameworks (Klappstein et al 2023, Pohle et al 2023), allow researchers to fit rich models to telemetry data in which movements may vary spatially and temporally as a function of environmental features (e.g., landcover types, distance to roads) and latent (unobserved) behavioral states (e.g., that may be associated with whether the individual has recently fed).…”

Accounting for behaviour in fine‐scale habitat selection: A case study highlighting methodological intricacies

“…The approach proposed by Nicosia et al (2017) would facilitate accounting for uncertainties in the HMM state decoding, but it would likewise prove difficult to incorporate the different sampling strategies investigated in this paper. This approach was recently extended by Klappstein et al (2023) to account for covariate influence on state transition probabilities, which may provide further insights into the interplay between behavioural state switching and habitat selection. With increasing availability of high‐resolution tracking data (sub‐minute to sub‐second resolution), SSFs may also be modified to define a ‘step’ as the straight‐line movement between turning points instead of regular positions determined by arbitrary GPS fix rates (Munden et al, 2021).…”