Case Study: Assessment of human-conditioned sorting behavior in dairy cows in farm research trials

Cow–human interactions influence and modulate group and individual behaviors of dairy cows. The objective was to test the effectiveness of human sorting on separating subgroups of dairy cows in on-farm studies and to assess the level of conditioning to this activity. Three sorting methods were compared: (1) human active sorting at the pen gate (AS); (2) human presence as passive sorting (PS); and (3) nonhuman gate sorting (GS). Holstein cows (n = 176) were randomly assigned to 1 of 2 subgroups (A = 91 animals and B = 85 animals) to be sequentially separated by 3 sorting methods. Each method was applied once per day after the morning milking during 5 d, and the proportions of cows correctly allocated in each group were recorded and compared. Additionally, an individual error index rate (number of errors per number of days observed) was calculated for each individual cow. When AS was applied, the total proportion of animals correctly sorted was 99.8%, whereas PS had 94.8% of total sorting accuracy. Cows lost the self-sorting behavior when exposed to nonhuman GS. The daily average of animals correctly placed was greater for AS when compared with PS (175 ± 1.7 vs. 166.6 ± 3.5; P = 0.005). Cows in the group that had longer walking distances to their subpen had greater individual error rate, especially when PS was applied. After a period of training, lactating dairy cows became conditioned to human sorting, which represents an opportunity to perform animal separation without intense human labor or stress for the animals.