Can stealing cows distort the results of feeding trials? An experiment for quantification and prevention of stealing feed by dairy cows from roughage intake control feeders

Practical experience suggests that stealing behaviour of dairy cows may affect the results in feeding trials, when automated feed troughs are used for measuring and controlling the feed intake. The objectives of the present study were to analyse the stealing behaviour from the log data of the feeder, quantify its biasing effects and make a proposal to solve the problem. Twenty dairy cows, ten of which were previously observed to steal roughage, were housed in a loose-housing system with 12 feed troughs (1.66 cows per trough). To induce the stealing behaviour, the animals were offered two kinds of feed, partial mixed ration (PMR) and silage (SIL). In a learning period (P0, three days), all the cows had free access to all the troughs. In periods 1 (P1, seven days) and 2 (P2, 14 days), 10 cows had access only to the SIL troughs and 10 cows only to the PMR troughs. In P2, barriers were built to prevent the stealing behaviour. In period 3 (P3, four days), the structure of the barriers was improved. We analysed two types of stealing behaviour: (1) stealing feed over a closed gate (S-GATE) and (2) stealing feed by displacing a legitimate cow from a trough (S-DISP). S-GATE by the SIL group, as measured by the average daily frequency and the amount of stolen feed (the PMR feed), decreased significantly (frequency: P = 0.001; dry matter: P = 0.001) from P1 (frequency: 8.0 ± 2.54; dry matter: 0.7 ± 0.26 kg; mean ± SEM) to P2 (1.0 ± 0.66; 0.1 ± 0.06 kg) and remained the same in P3 (1.0 ± 0.47; 0.1 ± 0.07 kg) as in P2. The frequencies of S-DISP, where the PMR feeding animals were displaced, was lower (P = 0.022) in P3 (0.0 ± 0.03 per animal per 24 hours) than in P1 (1.3 ± 0.60) and P2 (1.3 ± 0.53), but the differences were not statistically significant in the post hoc analyses. There were remarkable individual differences in the frequencies of how often a cow was detected to steal over a closed gate or being displaced from a trough. To conclude: stealing occurred frequently enough to bias the feed intake results in the experimental setup where cows clearly preferred one feed (PMR) over the other (SIL). However, appropriate barrier structures alleviated the problem markedly.