Devin S. Johnson, Joshua M. London, Mary-Anne Lea, and John W. Durban. 2008. Continuous-time correlated random walk model for animal telemetry data. Ecology 89:1208–1215.

Appendix C. Additional details for the analysis of marine mammal data.

Here we present some additional details and results for the analysis of the two marine mammal data sets, a harbor seal (Phoca vitulina), and a northern fur seal (Callorinus ursinus) pup. Figure C1 illustrates the data augmentation procedure for incorporating a covariate such as dry time to account for seal haul-out. Table C1 gives the parameter estimates for the CTCRW models for each data set. The CTCRW model was parameterized in terms of log values because all parameters ranges are strictly positive. Thus, the likelihood surface tends to be more well behaved and can be maximized over an unbounded space when log values are used. Confidence intervals for parameters on the original scale are given by back transformation of the confidence intervals on the log scale. Because the interval represents percentiles of a sampling distribution and exp() is a 1-to-1 function, the percentiles are preserved.

FIG. C1. Example of location data augmentation for harbor seal haul out model. Dry time data is a separate data set obtained from the telemetry device. The gray rows illustrate merging of the two data sets.

TABLE C1. MLEs for marine mammal movement example analysis.

†Parentheses include 95% asymptotic normal CI.