The objective of this study was to determine the most appropriate lactation curve representation, a 4th order Legendre polynomial (LP) function or a mechanistic mammary gland (MG) model, for various lactation curve shapes of grazing cattle in New Zealand. The ability of two optimisation methods, evolutionary algorithm (EA) and the Newton method, used to find parameter values that minimised mean prediction error (MPE) was also determined. The 95% confidence intervals for the parameters values and goodness of fitness for each lactation function and numerical algorithm were obtained with a bootstrapping strategy. Milk yields of three different cows were chosen to describe bi-peak, highly variable, and springpeak lactation curves. The MG function was able to find the best fit between predicted and actual values for the bi-peak and highly variable lactation curves withMPE of 2.7 to 3.7%, compared with MPE of 6.5 to 11.9% for the LP function applying Newton and EA. For the traditional lactation curve, the LP function applying Newton had the lowest MPE (1.7%) with the MG function applying an EA the highest MPE (4.1%). Overall, minimal differences in MPE were observed when solving using the Newton and EA. However, the Newton method fixed some parameters values of the MG function to achieve approximately the same MPE as the EA which manipulated all parameter values. The results of this study illustrate that a MG function solved with an EA is an accurate and efficient way to model non-standard lactation curves but a LP function in combination with the Newton method canbe better option for more standard lactation curves
Proceedings of the New Zealand Society of Animal Production, Volume 67, Wanaka, 209-214, 2007
|Download Full PDF||BibTEX Citation||Endnote Citation||Search the Proceedings|
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.