Estimating the optimal number of sampling days and patterns for recording calf behaviours in pre-weaning dairy calves

Abstract

Continuous video recording is the most accurate method for monitoring animal behaviours. However, a limited number of sampling days are often used to minimize the time and improve the efficiency of observation. To summarize the weekly behavioural data, most studies have utilized different sampling durations (sampling days) and patterns (non-consecutive vs. semi-consecutive vs. consecutive). It is still unknown what minimum sampling time and optimal combination patterns during a week are required to obtain accurate behavioural data in pre-weaning dairy calves. The aim of this study was to determine the minimum number of sampling days and optimal combination patterns of days for recording rumination, lying, standing, and performing non-nutritive oral behaviours of 6-week-old pre-weaning dairy calves. Behavioural data of 16 dairy calves was continuously recorded for a week (168 h/calf, wk 6 of age). The mean time spent performing each behaviour for 7 days was compared with data obtained from fewer sampling days ranging from 1 to 6 days with different combination patterns (e.g., consecutively: d 1–2–3; non-consecutively: d 1–3–5). The coefficient of determination between continuous behavioural data recording (for 7 days) and the different patterns with fewer recording days was calculated using the R statistical language. Combination patterns were considered accurate if they met 3 criteria: coefficient of determination ≥ 0.75, slope = 1, and intercept = 0. As expected, the strength of linear association between behavioural measures obtained from continuous recording (7 days) and fewer recording days increased as the sampling days increased. In terms of the R2 of specific sampling days with different combination patterns, the data from 4 or more sampling days were required to accurately estimate ruminating and performing of non-nutritive oral behaviours, which was significantly higher than the data from fewer sampling days. In comparison, data from 2 and 3 sampling days were required to accurately estimate lying and standing behaviours. Compared to consecutive patterns, a combination of non-consecutive days was more suitable for predicting the continuous data for rumination and performing non-nutritive oral behaviours. In some instances, a non-consecutive combination of even 3 sampling days was sufficient to effectively represent the time spent on rumination and performing non-nutritive oral behaviours (e.g., d 1–4–7). On the other hand, sampling consecutive (e.g., d 3–4) days was more suitable for accurately representing the time for standing behaviour than non-consecutive groups.