Using qualitative behaviour assessment (QBA) to explore the emotional state of horses and its association with human-animal relationship

This study aimed to apply qualitative behaviour assessment (QBA) to horses farmed in single boxes, in order to investigate their emotional state and explore its association with indicators of human-animal relationship. A fixed list of 13 QBA descriptive terms was determined. Three assessors experienced with horses and skilled in measuring animal behaviour underwent a common training period, consisting of a theoretical phase and a practical phase on farm. Their inter-observer reliability was tested on a live scoring of 95 single stabled horses. Principal Component Analysis (PCA) was conducted to analyse QBA scores and identify perceived patterns of horse expression, both for data obtained in the training phase and from the on-farm study. Given the good level of agreement reached in the training phase (Kendall W = 0.76 and 0.74 for PC1 and PC2 scores respectively), it was considered acceptable in the subsequent on-farm study to let these three observers each carry out QBA assessments on a sub-selection of a total of 355 sport and leisure horses, owned by 40 horse farms. Assessment took place immediately after entering the farms: assessors had never entered the farms before and were unaware of the different backgrounds of the farms. After concluding QBA scoring, the assessors further evaluated each horse with an avoidance distance test (AD) and a forced human approach test (FHA). A MANOVA test was used to assess the association of the AD and FHA tests with the on-farm QBA PC scores. The QBA approach described in this paper was feasible on farm and showed good acceptability by owners. In the analysis of on-farm QBA scores, the first Principal Component ranged from relaxed/at ease to uneasy/alarmed, the second Component ranged from curious/pushy to apathetic. Horses perceived as more relaxed/at ease with QBA showed less avoidance during the AD test (P = 0.0376), and responded less aggressively and fearfully to human presence in the FHA test (P < 0.0001). Our results support the hypothesis that QBA is sensitive to the quality of human contact in horses.