To conserve the mobility of species across changing land and seascapes, we must first understand how much space is necessary to maintain stable, interconnected populations. Home range estimation allows managers and policymakers to easily visualize the habitats most commonly used by species of conservation concern.
Home range estimation presents several quantitative challenges and is prone to statistical biases that can lead to underestimation (Fig. 1). This can have negative impacts on conservation outcomes as it may result in conservation managers underestimating how much land should be protected or overestimating the number of animals that a region can sustain.
Smithsonian scientists have overcome these biases by first accounting for the autocorrelation present in telemetry data. Autocorrelation is a statistic that contains the signature of the animal’s unknown movement path. As a result, autocorrelation-informed methods (autocorrelated kernel density estimation; AKDE) provide more reliable predictions of where animals may travel in the future as compared to traditional methods that depend on an assumption of unrelated locations (with no movement path connecting them; Fig. 2).
GPS location data are a discrete sequence of animal locations estimated at specific times and represent a sparse approximation of the animal’s unknown movement path. However, a signature of the underlying movement path may remain in the data via its autocorrelation structure. Autocorrelation in animal tracking data is an expression of the fact that locations obtained more closely in time also tend to geographically closer. The precise relationship between distance and time contains a wealth of information regarding the animal’s movement behavior and can be leveraged to make more accurate predictions.