Developing Methods for Calculating Individuality in Moose Movement Behavior from Remotely-Collected Location Dat

A rapidly-expanding literature on behavioral syndromes has revealed animal "personalities," or behavioral differences that are consistent through time and across contexts, in a wide range of taxa. Despite evidence of behavioral syndromes in many species, little has been done to consider the importance of this behavioral variation for management plans, which are developed based on averages. One problem is the intractability of conducting in-field behavioral assays on many managed wildlife populations. We studied the movement behavior of 35 moose (Alces alces) in northeastern Minnesota to determine if behavioral syndromes can be detected remotely in this population. If remote detection is possible, then behavioral variation could be assessed without the challenges associated with in-field assays. Location data were used to calculate various movement-related behavioral metrics, and landcover and terrain maps were used to quantify features of the environment. Behavioral metrics were used in a cluster analysis to look for patterns within and among individuals, and variation was also related to environmental variables. The optimal clustering approach included two metrics (daily path length and net daily displacement) to define five clusters (R2 = 0.94). When observations were plotted by NMDS, there was no distinct separation among the cluster groups. Clusters were not explained by the measured environmental variables, and there was a low rate of reclassification for individuals across multiple years. The methods were sufficient to produce general patterns consistent with what is known about moose behavior (e.g., reduced movement in winter, high use of forested cover types, and increased preference for aquatic features in summer). Overall, when using this approach it was unclear which variables were informative and appropriate for inclusion in clustering. The implications of behavioral variation for management remains an important subject, and we recommend conducting an in-field behavioral assay and then using the resulting behavioral scores to inform a multivariate analysis of movement-related metrics. This approach would help to determine which, if any, of the remotely-calculated metrics are able to identify behaviorally-defined groups.