Document Type

Article

Publication Date

2009

Abstract

This work aimed to apply a combined qualitative and quantitative approach to the interpretation of an on-farm behaviour test for horses, and to examine whether 1 month of handling would affect the response of yearlings to an unfamiliar stationary human in their home environment. Throughout a 1-month period, 14 Thoroughbred Yearlings (16 ± 0.22 months old) that had formerly experienced minimal contact with humans, were handled daily for about 45 min. The yearlings were tested twice, just before and just after the handling period. The behaviour of the horses during the tests was both video-recorded and directly recorded by the experimenter using an instantaneous time sampling recording method. Quantitative analysis of these data was achieved using principal component analysis (PCA). Qualitative analysis took place from video clips using a free choice profiling (FCP)methodology that requires observers to generate their own qualitative descriptors of behaviour, and in a second phase instructs these observers to quantify their personal descriptors on a Visual Analogue Scale. Observers were 21 veterinarians who were unaware that the horses had been handled in half of the clips and not in the other half. The data generated through FCP assessment were analysed using generalised procrustes analysis (GPA). Any differences in behaviour that may have occurred before and after the handling period were evaluated by comparing horse scores on the main PCA and GPA factors using a Wilcoxon matched-pairs test. To compare qualitative and quantitative assessments, both the quantitative behaviour measures and the qualitative behaviour scores were correlated to the main PCA factors obtained from the quantitative analysis using Spearman’s rank correlation. PCA analysis revealed three main factors (explaining 30%, 23% and 21% of the total variation between horses, respectively). The first factor showed high-negative loadings for immobile behaviour and high-positive loadings for contact and nibbling behaviour, and indicated that the horses tended to be more inclined to approach and contact the experimenter after handling (p = 0.08). GPA analysis revealed two main factors of expression (explaining 51.4% and 10.2%, respectively). Both factors indicated significant qualitative differences in the behavioural style of yearlings before and after handling (p < 0.05 and <0.01, respectively), characterising yearlings as ‘suspicious/nervous’ and ‘impatient/reactive’ before handling, and as ‘explorative/sociable’ and ‘calm/apathetic’ after handling. The correlation betweenGPAfactor1 scores with PCA factor 1 scores was highly significant (Spearman’s r = 0.75; p < 0.001), while those between GPA factor 2 scores with PCA factor 2 and 3 scores were not significant (r = -0.255; ns and r = 0.251; ns, respectively). On the whole a meaningful relationship was found to exist between the quantitative and qualitative behavioural assessments of the horses’ behaviour, indicating that these methods may be usefully combined in interpreting a behavioural test involving the presence of an unfamiliar human person.

Comments

This file contains a post-print version of the article, which has the same content as the final edited version but is not formatted according to the layout of the published journal.

Share

COinS