Conference on Applied Statistics in Agriculture

USING TIME-SERIES INTERVENTION ANALYSIS TO MODEL COW HEART RATE AFFECTED BY PROGRAMMED AUDIO AND ENVIRONMENTAL/PHYSIOLOGICAL CUES

This research is the first use of Box-Jenkins time-series models to describe changes in heart rate (HR) of free-ranging crossbred cows (Bos taurus) receiving both programmed audio cues from directional virtual fencing (DVFTM) devices and non-programmed environmental/physiological cues. The DVFTM device is designed to control the animal's location on the landscape. Polar Accurex® devices were used to capture HR every minute between 19 and 24 March 2003, when three mature free-ranging beef cows, previously habituated to the DVFTM device, were confined to a brush-infested area of an arid rangeland paddock. Global positioning system (GPS) electronics were used to record each cow's location approximately every minute while it was in a 58 ha virtual paddock (VPTM) and every second when it penetrated a virtual boundary (VBTM). The cows never escaped through the VBTM, although they penetrated it a total of 26 times in 11 different events, at which times they received programmed audio cues lasting from 1 to 56 s. Plots of these data reveal that HR spikes from programmed audio cues all fell within textbook range for cow HR (40-186 beats per minute, bpm). Heart rate spikes were analyzed using Box-Jenkins intervention analysis models, which showed that for both audio and selected environmental/physiological events, HR spikes typically returned to pre-cuing "baseline" levels in about one minute. However, the longest return-time to baseline (about four minutes) was for an environmental/physiological event of unknown origin. HR, animal location, weather and other electronic data should be measured at equally-spaced time intervals using a single time stamp to accurately associate HR changes with possible causes.