In vivo prediction of the carcass fatness using live body measurements in Pekin ducks
Fang-Bin Lin,Feng Zhu,Jin-Ping Hao,Fang-Xi Yang,Zhuo-Cheng Hou
DOI: https://doi.org/10.3382/ps/pey079
2018-07-01
Abstract:The purpose of this study was to evaluate the correlation between live body measurements and several fat traits in Pekin ducks, and ultimately to formulate multiple regression equations for the in vivo estimation of the carcass fatness of Pekin ducks. Several traits were measured in a total of 208 Pekin ducks aged 6 wk (107 males and 101 females). All ducks were weighed and measured for a set of body measurements including live body weight, body slope length, breast muscle thickness, skin fat thickness, chest width, keel length, and neck length. The breast muscle thickness and skin fat thickness was measured using B-scan sonography. Carcass information, including eviscerated weight, subcutaneous fat with skin weight, and abdominal fat weight, was collected after slaughter. Our results revealed that sex effects on most traits were significant (P < 0.05), and that the weight of subcutaneous fat with skin was significantly correlated with live body weight (r = 0.57 to 0.71, P < 0.01). Four additional traits of males were closely correlated with the weight of subcutaneous fat with skin, namely breast muscle thickness (r = 0.20, P < 0.01), skin fat thickness (r = 0.43, P < 0.01), chest width (r = 0.24, P < 0.01), and neck length (r = 0.20, P < 0.05). The abdominal fat weight, percentage of fat, and percentage of subcutaneous fat with skin of ducks were significantly correlated with live body weight (r = 0.38 to 0.43, P < 0.01), and skin fat thickness (r = 0.38 to 0.49, P < 0.01). These traits provided the basis for constructing regression equations to predict weight (or percentage) of subcutaneous fat and abdominal fat with high values of coefficients of multiple correlation (R) between the dependent variable and the independent variables. Two equations were verified to be applicable in other duck groups, with high accuracy, as more than 80% of estimated values were within the margin of error (<10%), compared with the actual values.