45418

Use of multivariate analysis to characterize the handling traits of stabled horses

Favorite this paper

The objective of this study was analyses the correspondence between handling quality traits of stabled horses. A questionnaire was applied to the managers of 30 horse stables in northwestern of S?o Paulo State, Brazil. The questions addressed the rearing conditions of three horses (randomly defined) in each of the stable horse (N = 90 horses), asking questions about the: frequency of concentrated feeding per day (CF: two or three), amount of hay per day (AH: < 7 kg, > 7 kg, ad libitum), site of hay feeding (HF: floor/feeder, bag/manger), frequency of single box cleaning (CL: once, twice or three times, >= 4 times per d), time spent on individual/collective paddock (PA: < 5 h individual paddock, < 5 h group paddock, > 5 h individual paddock, > 5 h group paddock), deworming (DE: absent or > 90 d, every 90 d), and teeth treatment (TE: absent, 180 d, 380 d, 720 d). Based on the answers a contingency matrix was constructed and each variable received a grade where the highest score corresponds to the most appropriate handling traits. Data were subjected to multiple correspondence analysis using the software Statistica 7(R) (P <= 0.01). The sum of eigenvalues of first two axes (x + y) represented 28.89% of the variation (13.23, 14.66%, respectively). Three correspondences groups were found the first one (named good traits) involving CL >= 4 times/d, PA > 5 h individual paddock, CF = three, and TE = 180 d; the second (bad traits) involving DE = absent and TE = 720 d, and the third (a mixed of good and bad traits), involving TE = 720 d, CL = once per d, and PA > 5 h group paddock. This result suggests that some horse stables are applying good practices of handling, while in other stables some handling practices should be improved. Moreover, the result showing the mixed group is suggestive of a lack of knowledge about horse handling. Because the sum of eigenvalues of first two axes (x + y) represented just 28.89% of the variation (13.23, 14.66%, respectively), more studies are needed to support results found in this study.