The use of differentially corrected global positioning system to monitor activities of cattle at pasture

Abstract

Global positioning system (GPS) technology is increasingly applied in livestock science to monitor pasture use and tracking routes, and is often combined with equipment for monitoring animal activity. As GPS data are referenced in time and space, it is hypothesised that parameters derived there from, such as distance travelled and aerial distance between the first and last point of a defined time interval, can be used to compute reliable estimates of daily activity budgets and hourly activity patterns of grazing animals. Fourteen Zebu cows grazing communal lands in semiarid western Niger were monitored on pasture throughout 1998. Observation of behaviour by man at intervals of 5 min was accompanied by logging geographical positions at intervals of 10 s using GPS units mounted on the animals. Thirty sets of combined observation and GPS data were randomly selected, stratified by season and split into a calibration and a validation data set of 15 individual itineraries each. Daily budgets and hourly patterns of the activities walking, grazing and resting did not differ significantly between the two data sets. Seasonal differences in activity patterns were similar in the two data sets. Among several parameters calculated from GPS data for 3 min time intervals, distance travelled, shortest distance between the first and the last logged position, average distance of each logged position from the first position and the ratio between the third and the first parameter proved to be the most reliable determinants for walking, grazing and resting. The linear discriminant functions derived from the calibration data correctly classified 79% of all observations in the calibration data set and 71% in the validation data set. Error count rates for resting, grazing and walking were 0.272, 0.195 and 0.176 for the calibration data and 0.419, 0.267 and 0.231 for the validation data. Grazing at low pace and resting were most difficult to differentiate due to the inherent location error of the GPS device. For the average grazing day of 540 min, the GPS-derived daily activity budget overestimated resting and walking time by 8 and 6 min, respectively, while grazing time was underestimated by 14 min for the validation data set. Evidence is provided that GPS devices with submeter accuracy will allow distinguishing more clearly between grazing at low pace and resting. The study demonstrates the potential to compute reasonable estimates of daily activity budgets and hourly activity patterns of grazing cattle from GPS recordings.