A Field-Based Time Geography for Wildlife Movement Analysis

Abstract

GIScience 2016 Short Paper Proceedings A Field-Based Time Geography for Wildlife Movement Analysis J. A. Long 1 School of Geography and Geosciences, University of St Andrews, Irvine Building, North Street, St Andrews, Fife, UK KY169AL Email: jed.long@st-andrews.ac.uk Abstract Field-based time geography is proposed as a new, specialized model for estimating wildlife utilization distributions (UDs) and home ranges. Field-based time geography represents the combining of classical time geography with least-cost path analysis. Here the derivation of field-based time geography is emphasized, paying particular attention to how it can be implemented in wildlife analysis. An example showing caribou movement in northern British Columbia is used to demonstrate field-based time geography and compare it with the Brownian bridge model, a popular method for delineating wildlife UDs. The results show how field-time geography is able to better represent the structure and barriers of the landscape, and provide alternative insights into wildlife space use. The entire process is implemented in R, making it attractive to movement ecologists. 1. Introduction One of the fundamental pieces of spatial information used in the study of wildlife movement is the home range. A home range is broadly defined as the area an animal uses in its normal day-to-day activities (Burt, 1943). From a GIScience perspective, a home range is a polygon, and thus represents this area discretely. Recognizing that animals typically do not use their home ranges evenly, alternatively a utilization distribution (UD) can be estimated, which represents the home range as a two dimensional probability density surface, where the values indicate the probability of observing the animal at different locations within (and outwith) the home range (Worton, 1989). Typically, a UD is represented as a raster grid after selecting an appropriate spatial resolution. After creating a UD it is common to estimate the home range simply as a % volume contour of the UD (most commonly the 95% volume contour). Many methods have been proposed for estimating UDs and home ranges and there is little consensus on which is best. Historically, kernel density estimation (Worton, 1989) has been the preferred approach, however in recent years Brownian bridges (Horne et al., 2007) have become increasingly popular. Time geographic methods have also been proposed as powerful alternatives for delineating home ranges (Long and Nelson, 2012) and UDs (Downs et al., 2011). However, one limitation that is present in all currently available home range and UD methods is the assumption that wildlife move within a homogeneous arena. That is, the underlying environment has no influence on the model used to calculate the home range and UD. By failing to consider how the underlying landscape characteristics (e.g., topography, structure, land cover) contribute to the formation of home ranges and UDs, current approaches fail to adequately capture how a heterogeneous environment contributes to observed movement patterns. In many cases, the natural environment constrains the area potentially visited by an animal, and thus home ranges and UDs are incorrectly estimated. To address this limitation, this paper proposes field-based time geography as a new tool for estimating wildlife UDs. Field-based time geography was originally proposed by Miller and Bridwell (2009) in the study of transportation, and thus focused on applications on a spatial network. Here, the conceptual framework outlined by Miller and Bridwell (2009) is extended to the study of wildlife movement in a two-dimensional field. In this context, field based time