Canonical discriminant analysis of body measures for prediction of body weight in horses of different breed and morphology

The estimation of horse body weight (BW) may represent an issue for the nutritionist when a scale is not available in stables. The adequacy of energy and nutrient content of the ration is calculated using BW as a variable. Modern horse breeds display huge morphological differences and alternative ways of BW prediction may display different extents of accuracy. This aspect still gathers the interest of the scientific community and prediction of BW by different means can lead to different outcomes. Also, body surface area (BSA) involved in different metabolic aspects for the sport horse covers an important role as to the nutritional management and body fluid distribution. Different scientific reports show complex calculations for BSA estimation based on morphometric measures, but the feasibility in field is low. In general, the more accurate BW estimation, the more reliable BSA. For this purpose, real BW (determined on digital scale) was comparatively examined with the estimated BW obtained by prediction equation (BWeq) and measuring tape (BWmt) for the horse. A total of 21 adult horses with a BCS of 4.5–5 (BCS 1–9 points scale) as pre-requisite for the enrolment in the trial, was weighed and assessed for the nutritional state. Different breeds were involved: French Trotter (2); Thoroughbred (4); Arabian (5); Saddle Italian (5); Saddle Belgian (3); Quarter horse (2). Two groups could be created as to meso-dolichomorph and dolichomorph breeds. Individual real BW was compared with estimated BWeq and BWmt by using a simple ANOVA model. Body measures used in the prediction equation for BW (wither’s height, heart girth and fore-leg girth) were collected on each horse and used to develop a canonical discriminant analysis (CDA) to ascertain if those measures are able to discriminate the two groups. Average BW, BWeq and BWmt displayed to statistically differ according to the morphology of animals (p<.001) but not as to methods of BW estimation (p=.057), in all cases being the prediction equation preferable to the measuring tape. In general, both estimation methods display to underestimate the real weight of the horse and this is more frequently observed in case of heavier horse breeds. CDA significantly separated the two groups and the best predicting measure appears to be the wither’s height. In conclusion, other body measures than heart girth or fore-leg girth may be considered when the rear of the horse may contribute to the differences between real and estimated BW.