|Title||Estimation of dairy goat body composition: A direct calibration and comparison of eight methods.|
|Publication Type||Journal Article|
|Year of Publication||2020|
|Authors||Lerch, S, De La Torre, A, Huau, C, Monziols, M, Xavier, C, Louis, L, Le Cozler, Y, Faverdin, P, Lamberton, P, Chery, I, Heimo, D, Loncke, C, Schmidely, P, Pires, JAA|
|Date Published||2020 Jun 27|
The objective was to compare eight methods for estimation of dairy goat body composition, by calibrating against chemical composition (water, lipid, protein, mineral and energy) measured post-mortem. The methods tested on 20 Alpine goats were body condition score (BCS), 3-dimension imaging (3D) automatic assessment of BCS or whole body scan, ultrasound, computer tomography (CT), adipose cell diameter, deuterium oxide dilution space (DOS) and bioelectrical impedance spectroscopy (BIS). Regressions were tested between predictive variates derived from the methods and empty body (EB) composition. The best equations for estimation of EB lipid mass included BW combined with i) perirenal adipose tissue mass and cell diameter (R = 0.95, residual standard deviation, rSD = 0.57 kg), ii) volume of fatty tissues measured by CT (R = 0.92, rSD = 0.76 kg), iii) DOS (R = 0.91, rSD = 0.85 kg), and iv) resistance at infinite frequency from BIS (R = 0.87, rSD = 1.09 kg). The DOS combined with BW provided the best equation for EB protein mass (R = 0.97, rSD = 0.17 kg), whereas BW alone provided a fair estimate (R = 0.92, rSD = 0.25 kg). Sternal BCS combined with BW provided good estimation of EB lipid and protein mass (R = 0.80 and 0.95, rSD = 1.27 and 0.22 kg, respectively). Compared to manual BCS, BCS by 3D slightly decreased the precision of the predictive equation for EB lipid (R = 0.74, rSD = 1.46 kg), and did not improve the estimation of EB protein compared with BW alone. Ultrasound measurements and whole body 3D imaging methods were not satisfactory estimators of body composition (R ≤ 0.40). Further developments in body composition techniques may contribute for high-throughput phenotyping of robustness.