Influences on the avoidance and approach behaviour of dairy goats towards an unfamiliar human—An on-farm study

• We evaluated four tests for assessment of the human-animal relationship in goats.
• Factors predicting goats reactions towards humans were identified by regression.
• Negative interactions during milking affected goats behaviour in all four tests.
• Relative importance of different factors varied depending on test type and measure.
• Avoidance tests (pen or feeding place) validly assess the human-animal relationship.

The human-animal relationship (HAR) is an important factor for successful animal husbandry and animal welfare. Thus, the HAR is included in on-farm assessments to evaluate overall welfare. For dairy goats, validated tests to assess the HAR are lacking. The aim of the present study was to evaluate several tests of the HAR for on-farm assessment regarding validity and feasibility. We visited 43 dairy farms in Austria and Germany and conducted one test with a stationary human (approach test in the pen—APP) and three tests with a moving human (avoidance test at the feed barrier—AvF, avoidance distance test at the feeding place—ADF, avoidance test in the pen—AvP). In addition, milker behaviour towards goats was observed, stockpeople’s attitudes towards interacting with goats were assessed via questionnaire and information was recorded about other potential influencing factors such as management, herd and housing characteristics. Convergent validity was investigated by (1) inter-test associations using a principal components analysis (PCA) and (2) by analysing associations of tests with potential influencing factors by use of multiple linear regression. Feasibility is described in terms of practicability of the tests in different farm designs and of time requirements. Variation between farms was high in all four tests. PCA revealed three factors including (1) all but one variable of the tests with a moving human, (2) all and only variables of the approach test, and (3) variables of AvF, AvP and APP where animals accepted touch or approached (nearly) to touch. Regression analyses revealed that the percentage of negative milker interactions (NegProportion) was the predominant influencing factor, being a predictor in seven of nine models: The higher NegProportion, the more avoidance and the less approach behaviour towards the unfamiliar test person were observed in all four tests (P ≤ 0.001 in 4 models, P ≤ 0.01 in one, P ≤ 0.05 in one and P = 0.16 in one model with a beta of |0.3| to |0.8|). Measures of AvF and AvP were explained best by human behaviour and were hardly influenced by confounding factors with an explained variance up to 80%. While feasibility of AvP was high, AvF and ADF could not be performed on about 30% of the farms due to unsuitable feeding equipment. However, AvF can be modified to improve feasibility. Avoidance tests at the feeding place and in the pen can validly assess the human-animal relationship on dairy goat farms as shown by the strength of inter-test associations and of the association with human-animal contact variables. A combination of avoidance tests at the feeding barrier and in the pen is recommended.