Test-retest reliability of buffalo response to humans.

This study aims to assess the test-retest reliability of buffalo response to humans and the relationship between the behaviour of stockpeople and buffalo. The research was carried out in 27 buffalo farms located in southern Italy. Milk production was recorded only in the 14 farms enrolled in the milk recording scheme. Before assessment informed consent was obtained by stockpeople. Human–animal relationship was assessed through two different tests: behavioural observations of stockperson and animals during milking and avoidance distance at the manger. These tests were repeated within one month to assess test–retest reliability. Stockpeople attitude was evaluated using a questionnaire divided into four sections (general beliefs about buffaloes, general beliefs about working with buffaloes, behavioural intentions with respect to interacting with buffaloes and job satisfaction) including 21 statements. A high degree of test–retest reliability was observed for all the variables concerning the behaviour of stockpeople and animals. The values of Spearman rank correlation coefficient (rs) ranged from 0.578 (P<0.01) for the number of kicks performed by the animals during milking to 0.937 (P<0.001) for the percentage of animals that move when approached by ≤0.5 m. Overall the coefficients of the variables measured on the stockpeople were lower than those obtained for the animal-based variables. A negative correlation between job satisfaction of stockpeople and number of steps performed by the animals during milking was found (rs=-0.372, P<0.05). During milking negative interactions performed by stockpeople both in absolute number and in percentage terms positively correlated with the number of kicks performed by the animals (rs=0.421 and rs=0.430, P<0.05; respectively). A positive correlation was also found between the number of negative interactions and the percentage of buffaloes treated with oxytocin (rs=0.424, P<0.05). Milk yield expressed as kg of milk/buffalo/year was positively correlated with the number of positive interactions (rs=0.588, P<0.01) and negatively correlated with the number of steps performed by the animals during milking (rs=-0.820, P<0.001). Finally, the higher was the mean avoidance distance at manger the larger was the percentage of animals treated with oxytocin (rs=0.387, P