The objective of the present study was to use bivariate mixture models to study the relationships between body weight (BW) and ascites indicator traits. Existing data were used from an experiment in which birds were housed in 2 groups under different climate conditions. In the first group, BW, the ratio of right ventricular weight to total ventricular weight (RV:TV), and hematocrit value (HCT) were measured in 4,202 broilers under cold conditions; in the second group, the same traits were measured in 795 birds under normal temperature conditions. Cold-stress conditions were applied to identify individuals that were susceptible to ascites. The RV:TV and HCT were approximately normally distributed under normal temperature conditions, whereas the distributions of these traits were skewed under cold temperature conditions, suggesting different underlying distributions. Fitting a bivariate mixture model to the observations showed that there was only one homogeneous population for ascites traits under normal temperature conditions, whereas there was a mixture of (2) distributions under cold conditions. One distribution contained nonascitic birds and the other distribution contained ascitic birds. In the distribution of nonascitic birds, the inferred phenotypic correlations (phenotypic correlations with 2 distinguishing underlying distributions) of BW with RV:TV and HCT were close to zero (0.10 and -0.07, respectively), whereas in the distribution of ascitic birds, the inferred phenotypic correlations of BW with RV:TV and HCT were negative (-0.39 and -0.4, respectively). The negative inferred correlations of BW with RV:TV and HCT in the distribution of ascitic birds resulted in negative overall correlations (correlations without 2 distinguishing distributions) of BW with RV:TV (-0.30) and HCT (-0.37) under cold conditions. The present results indicate that the overall correlations between BW and ascites traits are dependent on the relative frequency of ascitic and nonascitic birds in the population.

Keywords