Combined analysis of multiple linear regression and principal components for predicting performance indicators in broiler chickens under commercial conditions

Abstract

This study aimed to assess the influence of biological, environmental, and management factors on broiler performance traits under commercial conditions using multiple linear regression and principal component analysis (PCA). We used a database containing 4,842 records of commercial broiler flocks (Ross 95 and Cobb 500) collected between 2019 and 2022, which included traits such as breeder ages, sex of flocks, type of housing (conventional or climate-controlled houses), stocking densities, placement seasons, litter reutilization, and slaughter ages. Performance responses included body weight (BW) and feed intake at different ages, daily weight gain (DWG), feed conversion ratio (FCR), and production efficiency index (PEI). Predictors were selected by stepwise backward regression using the Akaike information criterion. Multicollinearity was assessed using variance inflation factors, and model was validated using 10-fold cross-validation. The PCA was applied to identify performance patterns across categorical groupings and components retained by Kaiser criterion with explained variance ≥ 60%. Males showed increases of 0.352 kg in BW at 42 days and 27.6 points in PEI compared to females, while broilers from young breeders reduced these values by 0.030 kg and 5.6 points, respectively. Each additional bird/m² increased BW by 0.050 kg and PEI by 7.6 points. Slaughter after 42 days decreased PEI by 5.4 points per day. Summer placements increased BW by 0.107 kg and PEI by 25.2 points compared to winter, while conventional houses in spring and summer led to performance reductions. Feed intake and DWG were influenced by sex of flocks, breeder age, stocking density, and their interactions with season and type of housing. The PCA explained 70.2% of the total variance, with PC1 strongly associated with BW, DWG, and PEI, and negatively with FCR. In conclusion, male and mixed-sex flocks showed superior performance, while broilers from young and second-cycle breeders had lower performance. Intermediate stocking densities (12-14 birds/m²) and earlier slaughter age improved outcomes. Fall and summer placements enhanced performance, although conventional housing under these conditions was detrimental.