Interactions among individuals (both intraspecific and interspecific) are of critical importance for their survival and for the use of resources. These interactions might vary depending on different factors, such as distance between individuals, sex, time of day or a particular season. To analyse these interactions, we develop a method to evaluate whether or not identified approach and retreat behaviours in pairs of individuals are statistically significant from random. Our method is based on calculating the angle between the absolute heading of one individual and the direction vector pointing towards the other in the pair. We then bin these angles based on the distance between the pair, and use the binned groups to characterize distance-dependent behaviour. By considering heading differences and the relative movement of each individual at each point in time, we extract various behaviours, such as following or side-by-side movement and then compare their frequencies of occurrence at different times of the day and in different seasons.
Results/Conclusions
We applied our methodology to GPS relocation data from 39 African elephants, collected between 2008 and 2015 in Etosha National Park, Namibia. The intervals at which these data were recorded varied among 4, 3 and 2 points per hour (10, 15, and 14 individuals respectively). Data were collected over different periods, ranging from 2.5 months to 4.5 years. From these data, we identified 136 pairs that had matching frequencies within overlapping of time periods and were close enough to one another for our method to yield interpretable results. Specifically, we identified and binned the frequencies of approach and retreat behaviours in relation to the pair distance: we used 9 different size classes, starting with a 0-50m bin and ending with a 5-10km bin. We also compared these frequencies across diurnal and seasonal cycles and as a function of the sexual state of the pairs (i.e. male-male, female-male, and female-female).