Can Movement Speed Predict Habitat Preference? Assessing the Influence of Topography, Village Proximity, and Land Cover on Forest Elephant (Loxodonta cyclotis) Movement and Preferences in Gabon.

Understanding animal habitat preference is crucial for the management of animal populations and planning of protected areas. However, current models for estimating habitat preference require arbitrary estimates of habitat availability, which introduce a level of uncertainty and qualitative inference that could affect model accuracy. To overcome this problem, Dickson et al. (2005) suggested that movement speed could be used as a metric of habitat preference, assuming that movement speed would be negatively related to habitat preference. However, this speed - preference model ignores potential changes in movement related to behavioral shifts or variations in terrain. To assess the generalizability and practicality of the speed-preference model, I examined the hourly movements of 56 GPS collared forest elephants (Loxodonta cyclotis) in Gabon, central Africa between 2015 and 2018 in the context of three relevant environmental covariates (land covers, topography, and village proximity). I analyzed changes in movement speed by attributing a single value for specific environmental characteristics to each movement step and estimated preferences by calculating the density of each individual’s GPS points within the covariate of interest from the individual’s travel range. I then modeled the relationship between speed and preference with a linear mixed model. Speed failed to predict preferences for different land cover types, and relationships between speed and preference for gradients of topography and village proximity changed in both direction and intensity. Therefore, although using speed to predict habitat preference avoids the limitations of other habitat preference models, this method requires further research to assess the complex interactions between speed and environmental covariates for different animal species.