Application of Non-linear Models to Egg Production Curves in Chickens.

Four mathematical models (compartmental (C), modified compartmental (MC), Wood (W), and Adams-Bell (AB)) were fitted to different patterns of egg production data in laying hens to compare the fits of these models.The egg production data were obtained from the two lines which were founded on a basis of divergent selection for egg quality traits. Egg production data of individual hens were classified in six patterns. The average hen-day rates of lay were calculated for every 10 days from the first egg for each pattern. The comparison of fits was based on the R2 adjusted for degrees of freedom and Akaike's information criterion (AIC).The fits of C, MC and AB models were similar for the egg production data showing a general pattern of curve. However, the model parameters could not be estimated by these three models for the data whose patterns showed abrupt decreases after the peak of egg production. The data in which the period from the first egg to the peak of egg production was short and linearly decreased after reaching the peak showed poor fits of the models. The parameters of all patterns could be estimated by the W model, but the fits were lower than those of the others.