Phase-dependent nutrient requirements of broiler breeder hens for day-old chick production in a field study using...

Day-old chick output from broiler breeder flocks is a key determinant of efficiency in the global poultry industry and is highly sensitive to breeder nutrition. This study aimed to quantify phase-dependent requirements of metabolizable energy and digestible amino acids (lysine, methionine, and threonine) using a large-scale commercial dataset and an interpretable modeling framework. Data from 2,685 flock-week observations (2016-2022) were analyzed across three production phases: early (26-32 weeks), peak (33-50 weeks), and late (51-62 weeks). Nutrient-response relationships were modeled using both classical regression and a single-neuron artificial neural network with a hyperbolic tangent activation, enabling flexible, nonlinear dose-response estimation while preserving biological interpretability. Nutrient requirements were defined as the intake required to achieve 95% of the asymptotic response and were estimated with bootstrap-derived confidence intervals. The ME requirements were highest in early production (453 kcal/d) and declined slightly during peak (449 kcal/d) and late production (448 kcal/d); the narrow numerical range across phases warrants cautious interpretation despite statistical significance. In contrast, amino acid requirements followed reproductive intensity, peaking during mid-production (digestible lysine 1.03 g/d, methionine 0.68 g/d, threonine 0.80 g/d) and decreasing thereafter. The nonlinear modeling approach consistently captured saturating responses that were not adequately described by linear models. These results demonstrate that nutrient requirements in broiler breeders are phase-dependent and can be effectively quantified using interpretable machine learning applied to field data. The proposed framework provides a practical tool for deriving biologically meaningful requirement estimates and supports more precise, phase-specific feeding strategies in commercial production systems.