Are mixed models more effective than linear models in predicting energy utilization for laying hens?

Abstract

This study was conducted to develop mathematical models that can overcome the limitations of linear energy partitioning models in laying hens comparing the metabolizable and net energy bases. Three linear models and 2 nonlinear mixed-effect models were fitted to predict the energy utilization efficiency for body tissue deposition and egg production, as well as the maintenance requirements for metabolizable energy (MEm) and net energy (NEm). A total of 30 diets were individually formulated to achieve values below and above the nutritional recommendation of the genetic guidelines. Heat production and energy metabolism were measured in laying hens during the production phase using 6 open-circuit respiration chambers. Diets were randomly assigned into 5 groups, and each diet was replicated 4 times with 6 hens per replication. The experimental protocol included a 5-d adaptation to the experimental diets, 2-d chamber adaptation, 4-d measurements of gas exchange, total excreta collection and growth performance recording under feeding conditions, and 1-d gas exchange measurement under fasting conditions. Feed intake, body weight, egg production, egg mass, and excreta production were measured daily. The variables for total heat production (THP) and fasting heat production (FHP) were obtained through the Brouwer equation using VO₂ and VCO₂ measurements. Apparent metabolizable energy intake (AMEi) was determined by the difference between the gross energy intake and excreted. Retained energy (RE) was determined as the difference between AMEi and THP. Subsequently, the RE was partitioned between RE in body and egg. The RE in egg and body were partitioned into RE as protein and fat. Statistical analyses involved linear regressions and nonlinear mixed-effect regressions of the main variables described. The values obtained for MEm, NEm, and the efficiency of energy retention in body (k_body), and egg (k_egg) in the linear models were 106 kcal/kg^0.75.d, 90 kcal/kg^0.75.d, 0.843, and 0,779, respectively. For nonlinear models the values were 94.15 kcal/kg^0.75.d, 81.72 kcal/kg^0.75.d, 0.463, and 0.638 for MEm, NEm, k_body and k_egg, respectively. Nonlinear models were considered the ideal choice to determine MEm and NEm requirements due to the lowest error. The non-linear mixed model provides a more accurate representation of energy utilization in laying hens compared to linear models, offering a better characterization of energy partitioning within the net energy bases.