ABSTRACT
Assessing feeding behavior is important in understanding the effects of nutrition and management on the well-being of dairy cows. Historically, collection of these data from cows fed with a Calan Broadbent Feeding System (American Calan Inc., Northwood, NH) required the labor-intensive practices of direct observation or video review. The objective of this study was to evaluate the agreement between the output of a HOBO change-of-state data logger (Onset Computer Corp., Bourne, MA), mounted to the door shell and latch plate, and video data summarized with continuous sampling. Data (number of feed bin visits per day and feeding time in minutes per day) were recorded with both methods from 26 lactating cows and 10 nonlactating cows for 3 d per cow (n=108). The agreement of the data logger and video methods was evaluated using the REG procedure of SAS to compare the mean response of the methods against the difference between the methods. The maximum allowable difference (MAD) was set at ±3 for bin visits and ±20 min for feeding time. Ranges for feed bin visits (2 to 140 per d) and feeding time (28 to 267 min/d) were established from video data. Using the complete data set, agreement was partially established between the data logger and video methods for feed bin visits, but not established for feeding time. The complete data set generated by the data logger was screened to remove visits of a duration ≤3 s, reflecting a cow unable to enter a feed bin (representing 7% of all data) and ≥5,400 s, reflecting a failure of the data logger to align properly with its corresponding magnetic field (representing <1% of all data). Using the resulting screened data set, agreement was established for feed bin visits and feeding time. For bin visits, 4% of the data was beyond the MAD. For feeding time, 3% of the data was beyond the MAD and 74% of the data was ±1 min. The insignificant P-value, low coefficient of determination, and concentration of the data within the MAD indicate the agreement of the change-of-state data logger and video data. This suggests the usage of a change-of-state data logger to assess the feeding behavior of cows feeding from a Calan Broadbent Feeding System is appropriate. Use of the screening criteria for data analysis is recommended.
