Optimal behaviour sampling and autocorrelation curve: modelling data of farmed foxes

Behavioural investigations frequently make use of video recordings. In a typical study, behaviour is recorded during a 24-h period using instantaneous sampling, a method whereby consecutive samples taken from the same animal are autocorrelated. This study sought to model the autocorrelation function and to evaluate the optimal sampling interval for recording periods of different length in farmed blue foxes. The data were based on video recordings. The autocorrelation was found to have two components: short-range and long-range. The range of the short-range component was from 2 to 4 min, implying that an animal continues the same operation without interruption for several minutes. The duration of the long-range component was from 20 to 75 min and was affected by the animals mental state. The optimal sampling interval was not the same for all the behaviours studied and, moreover, depended on the length of the recording period. If the recording period lasted for 3 to 5 days, the optimal sampling interval was from 10 to 15 min; if, however, there was only one recording day, the optimal sampling interval was from 4 to 6 min.