Outlying data was removed in feeding activity (5) and locomotion (2),
identified by values outside the range of the mean to which two standard
deviations were added or subtracted, to remove data points
whose residuals were in the top or bottom 2.5% of the distribution
[31]