Set up of a sampling strategy for the collection of animal-based welfare indicators during milking

We hypothesised milking time to be the period of the production process that allows observers to have closer contact with dairy goats, hence being the ideal location/time to assess certain animal-based welfare indicators. However, for the development of a practical welfare assessment scheme, observing all the animals during milking would be too time consuming. The aim of the present study was to define a sampling strategy that can reduce the observation time and at the same time produce a valid and unbiased result. We observed milking in two farms (Farm1: 303 goats; Farm2: 141 goats). In each farm, goats were housed in two pens, so we had a total of four pens (191 and 112 goats each in Farm1; 65 and 76 goats each in Farm2). We collected data on lameness, body condition score, cleanliness, teat and udder abnormality and overgrown claws. Data were compared by chisq test. We found significant differences (at least P<0.01) between farms for all the observed variables, except for lameness (P=0.08). Within each farm, no differences between pens were recorded. A minimum of 3 to a maximum of 6 milking groups were formed from each pen, depending on the size of the milking parlour. We compared the prevalence of each welfare indicator in the whole pen with its prevalence in each milking group. In most cases, each single milking group did not statistically differ from the whole pen, and the central groups were able to better reflect the welfare situation of the whole pen. However, a clear and often significant increase of the prevalence of lameness was observed in the last milking groups. Our preliminary results suggest that a reliable sampling strategy can be used during milking to gather information about the welfare condition of dairy goats, thus reducing the time needed for data collection. When more than one pen is present in a farm, it seems that one pen can give sufficient information to represent the whole farm welfare level for the considered variables, provided that observations are carried out on the central milking groups. This seems particularly important for lameness. Further observations in more farms are required in order to confirm this hypothesis.