Which measures of acceleration best estimate the duration of locomotor play by dairy calves?

Abstract

Measures of acceleration have been used as automated measures of the locomotor play of calves. We examined which measures of acceleration were best correlated with locomotor play and how the sampling rate of the accelerometer affected the correlations. Accelerometers were attached to 30 6–7 week-old Holstein calves, each of whom was released into a large arena for 10min on each of 10d. The calves were kept with their dams for four trials (baseline), partially separated from their dam (physically separated but in visual contact) for three trials and totally separated from their dams for three trials. We correlated measures of acceleration of the leg (summed acceleration, and the number of peaks in acceleration in each axis and overall, sampled at 33Hz, 11Hz and 1Hz) with the duration of running and walking, and the frequency of jump/kick (scored from video-recordings). Peaks in acceleration occurred most often when the calves were running. When all trials (n=271) were used, the overall number of peaks in acceleration measured at 33Hz had the largest correlation with the duration of running (r=0.96), and had large correlations with the frequency of jump/kick (r=0.86) and the ratio of locomotor play (the duration of running and jump/kick combined) to walking (r=0.75). However, a sizeable correlation (r=0.90) was also found between the duration of running and the duration of peaks in the vertical direction at a 1Hz sampling rate, which with the type of accelerometer used here, would allow measures of acceleration to be taken for 18h. During the first trials after both partial separation and total separation, the change in the number of peaks in acceleration was significantly correlated across calves with the change in duration of running (r=0.90; n=24; P<0.001 after partial separation; r=0.97; n=28; P<0.001 after total separation). Both the duration of running and the peaks in acceleration after both partial and total separation were correlated to a similar extent (P<0.001) with the digestible energy intake of the calves. Accelerometers provide good estimates of the locomotor play of calves, can distinguish this from walking and can show the effects of reduced energy intake on locomotor play. Reducing the sampling rate and measuring acceleration in only the vertical axis greatly lengthens the time that measures can be taken with this type of accelerometer without markedly reducing the accuracy of the estimates.