DATA AVAILABILITY STATEMENT : The data are available at: https://osf.io/v5pnc/
SUPPLEMENTARY MATERIAL : S1 Appendix. Calculation of average travelled distance using coefficient estimates associated to step length.
https://doi.org/10.1371/journal.pone.0272538.s001
S1 Table. Values and definition [from c] of model parameters used to simulate multi-state correlated random walks in three scenarios of landscape patchiness.
https://doi.org/10.1371/journal.pone.0272538.s002
S2 Table. Coefficient estimates along with their 95% confidence interval (95% CI) of the mixed-effects generalized linear model with binomial distribution (HMM-SSF + GLMM) and the multi-state correlated random walk model (HMM-CRW) to predict probability of switching from encamped to travelling mode, in 500 simulated foragers moving among resource patches and avoiding a predator.
In resource patch is a dummy variable indicating whether the forager is within a resource patch (i.e., patch quality >0), equals the actual distance of the predator from the forager (dPredator) when dPredator ≤ 0.8 km and 0.8 km, otherwise. log(dPredator) is the natural logarithm of dPredator.
https://doi.org/10.1371/journal.pone.0272538.s003
S3 Table. Coefficient estimates along with their 95% confidence interval (95% CI) of mixed-effects generalized linear models with binomial distribution to predict probability of switching from encamped to travelling mode of movement, in plains bison during summer in Prince Albert National Park (SK, Canada).
Each table represents estimates for a specific threshold probability (Pthreshold) used to categorized transition and non-transition from the conditional probabilities of being in encamped or travelling state, obtained from the fit of the HMM-SSF to plains bison trajectories. was set to the actual distance between bison and wolf (dwolf) when dwolf≤dthreshold and dthreshold, otherwise.
https://doi.org/10.1371/journal.pone.0272538.s004
S1 Fig. Simulated heterogeneous landscape used in the multi-state biased correlated random walk simulations, from gaussian random field with an exponential covariance function with variance = 1, nugget = 0 and a set of patch concentration (μQ) and patch size (γQ) resulting in three level of patchiness: 1) low (μQ = -1.5, γQ = 2), 2) intermediate (μQ = -0.5, γQ = 2) and 3) high (μQ = 1, γQ = 10).
https://doi.org/10.1371/journal.pone.0272538.s005
S2 Fig. Distribution of distance to the closest waterhole according to the mode of movement estimated from the HMM-SSF for 18 zebras in Hwange National Park during the dry hot season.
The conditional probabilities of being in each state, obtained from the fit of the HMM-SFF, were dichotomized to 0–1 based on a 0.5 threshold to determine the state of the individual at each step on its trajectory.
https://doi.org/10.1371/journal.pone.0272538.s006
S3 Fig. Log-likelihood profile from mixed-effects generalized linear model with binomial distribution to predict probability of switching from encamped to travelling mode of movement, according to a gradient of threshold distance, dthreshold.
https://doi.org/10.1371/journal.pone.0272538.s007
S4 Fig. Total number of switches from encamped to travelling mode of movement according to day time, estimated using conditional probabilities of being in each state, obtained from the fit of the HMM-SFF to plains bison trajectories followed during the summers 2005–2016.
We then separated the day in four periods: Night: 22:00–02:00, Dawn: 03:00–06:00, Day: 07:00–15:00 and Dusk: 16:00–21:00.
https://doi.org/10.1371/journal.pone.0272538.s008