Caractérisation morpho-pondérale chez le poulet local et prédiction du poids vif par des techniques d’exploration des données

Abstract

Local chicken breeds are an iessential tool for food security and the fight against poverty for rural families. This study consists initially of the phenotypic characterization of local chickens in the wilaya of Mostaganem, and to establish models for the prediction of the live weight of the chicken from certain body measurements. One hundred chickens (50 males and 50 females) were used in the current study. The average chicken weight in the study area was 1307,97±186,8g. Adult animals were mainly of medium which is reflected by body length and wingspan of 28,75±1,22 cm, and 41,05±4,74 cm, respectively. he phenotypic correlations between the quantitative parameters studied vary between -0,017 and +0,96. The live body weight wassignificantly correlated to crest height (+0,82, P<0.01) and wingspan (+0,78; P<0,001). The goodness-of-fit test was performed to compare the predictive performance of the artificial neural network (ANN) and automatic linear regression (RLA) to predict bird live body weigh. The coefficient of determination (R2) values were 0.973 and 0.924 for RNA and RLA, respectively. These results deserve to be confirmed for the implementation of selection programs and strategies to improve the production potential of local chickens.