Weight gain responses of laying-type pullets to methionine plus cystine intake

Abstract

Each individual in the population has a distinct maximum growth potential, and the growth curve may vary depending on the response to nutrient intake, growth phase and variability among animals. The present study aimed to (1) model weight gain (WG) response to methionine+cystine (Met+Cys) supply using different mathematical functions, (2) identify functions that better fit the growth responses of pullets, (3) determine the Met+Cys requirements that maximize WG based on breeding standards and (4) partition the Met+Cys requirements for WG and maintenance. Three trials were performed using 1448 laying-type pullets. We adopted a completely randomized design with eight treatments and six replicates. The first trial (2 to 6 weeks, P1) used 15 pullets per experimental unit. The second and third trials (8 to 12 weeks, P2; 14 to 18 weeks, P3) were used eight pullets per replicate. The Met+Cys levels were obtained using a dilution technique. The mathematical functions used to describe WG responses to Met+Cys intake were broken line, broken line with curvilinear ascendancy, Michaelis–Menten, saturation kinetics and three logistic and three exponential models. Models were selected using the Bayesian information criterion and evaluated by residual analysis. It was possible to model the responses using the studied functions. The best functions were obtained by logistic and sigmoidal models in P1 and P2, and with the broken line by the curvilinear ascendancy model in P3. The Met+Cys intake that determined the maximum potential for WG (WGmax) in P1, P2 and P3 were 313, 381 and 318 mg/day, respectively. The Met+Cys requirements for WG were 20, 22 and 27 mg/g, and for maintenance were 214, 53 and 30 mg/kgBW0.75 for P1, P2 and P3, respectively.

Modeling the response of Japanese quail to arginine intake

“…The broken-line model approach commonly yields lower requirements than non-linear models ( Fatufe & Rodehutscord, 2005 ). All values obtained for R eq were based on area, limiting the use of the broken-line model to define the breakpoint, and according to Silva et al (2020) , this point is similar between linear and non-linear models.…”

“…Therefore, the value of b 1 ≠ 0 indicates the model’s prediction bias ( St-Pierre, 2003 ; Silva et al, 2020 ). The b 0 value indicates the general error and is related to the scale of the difference ( St-Pierre, 2003 ; Silva et al, 2020 ). The precision value of the model was calculated by considering the 1 − R 2 adj ( St-Pierre, 2003 ; Silva et al, 2020 ).…”

“…The variables arginine intake ( X ) and arginine deposition in eggs ( Y ) were standardized according to the unit system per kilogram of metabolic weight (BW 0.67 ), converting them into mg/kg BW 0.67 , according to Silva et al (2020) . Two models were used to interpret the relationship between Y and X , a monomolecular function was modified for four parameters adapted from Kebreab et al (2008) and the saturation kinetics model ( Mercer, 1982 ).…”

“…The decision rule assumed that the model was impartial as correlation approaches 1 and when R 2 adj approaches 0, i.e., the residuals are not correlated with the predictions. The slope value, b 1 , as a function of y i , must tend towards zero for the model to be impartial ( St-Pierre, 2003 ; Silva et al, 2020 ). Therefore, the value of b 1 ≠ 0 indicates the model’s prediction bias ( St-Pierre, 2003 ; Silva et al, 2020 ).…”

“…The slope value, b 1 , as a function of y i , must tend towards zero for the model to be impartial ( St-Pierre, 2003 ; Silva et al, 2020 ). Therefore, the value of b 1 ≠ 0 indicates the model’s prediction bias ( St-Pierre, 2003 ; Silva et al, 2020 ). The b 0 value indicates the general error and is related to the scale of the difference ( St-Pierre, 2003 ; Silva et al, 2020 ).…”