Technical note: Recording rules for behavioral studies in growing heifers fed high-concentrate diets.

Continuous recording over 24 h is an accurate method for behavioral measurements in ruminants but is very time-consuming. The aim of this study was to compare different recording rules, in particular continuous recording (CR) with different time samplings (TS) methods, when the behavior of growing heifers housed individually and fed a high-concentrate diet is recorded. Behavior of 8 growing heifers was recorded on 6 non-consecutive days. Continuous recording over 24 h, considered the control treatment, was used to evaluate TS methods by recording all the behaviors that occurred for 60 s at intervals of 2 (S2), 5 (S5), 10 (S10), 15 (S15), 20 (S20) and 30 (S30) min. Three additional TS methods were considered: a sampling of the first 10 min of every h (FS10), and a sampling during 4 h (4PF) and 8 h post-feeding (8PF). Two procedures to assess the accuracy of the relationship between CR and TS methods were used: a) data analysis using the GLIMMIX procedure of SAS together with the Pearson correlation coefficient, and b) linear regression analysis. In this last case, 1-time sampling was considered to accurately reflect true values obtained with CR if the following 3 criteria were met: the coefficient of determination () was greater than 0.76, the intercept did not differ from 0 ( > 0.05), and the slope did not differ from 1 ( > 0.05). The results of the present study suggest that the latter approach provides a much more focused and reliable way of ensuring the accuracy of each TS method. The S5 TS method produced an accurate estimate for all the behavioral activities, except for drinking, social behavior and rummaging in wood shavings, where it would be necessary to consider the S2 TS method, and S10 for self-grooming. Ruminating and resting can be accurately estimated using the S20 and S30 TS methods, respectively. However, FS10, 4PF and 8PF were inadequate TS methods to measure all the activities studied in the present experiment. In conclusion, not all the TS methods tested are adequate for measuring the behaviors recorded.