Abstract
The objective of this study was to evaluate four mathematical models with regards to their fit to lactation curves of Holstein cows from herds raised in the southwestern region of the state of Parana, Brazil. Initially, 42,281 milk production records from 2005 to 2011 were obtained from "Associação Paranaense de Criadores de Bovinos da Raça Holandesa (APCBRH)". Data lacking dates of drying and total milk production at 305 days of lactation were excluded, resulting in a remaining 15,142 records corresponding to 2,441 Holstein cows. Data were sorted according to the parity order (ranging from one to six), and within each parity order the animals were divided into quartiles (Q25%, Q50%, Q75% and Q100%) corresponding to 305-day lactation yield. Within each parity order, for each quartile, four mathematical models were adjusted, two of which were predominantly empirical (Brody and Wood) whereas the other two presented more mechanistic characteristics (models Dijkstra and Pollott). The quality of fit was evaluated by the corrected Akaike information criterion. The Wood model showed the best fit in almost all evaluated situations and, therefore, may be considered as the most suitable model to describe, at least empirically, the lactation curves of Holstein cows raised in Southwestern Parana.
Highlights
The lactation curve is a graphic representation of the milk production from an animal throughout a defined period (Yadav et al 1977)
A particular situation is the evaluation of the model of Brody et al (1924), which has, so far, been described erroneously and certainly invalidates the results described in literature
The data were ranked according to the order of parity of the cows and within each parity order, the animals were divided into quartiles (Q25%, Q50%, Q75% and Q100%) according to total milk yield at 305 days
Summary
The lactation curve is a graphic representation of the milk production from an animal throughout a defined period (Yadav et al 1977). The wide use of this model is, probably due to its ability to fit relatively well to a broad kind of data such as animals of different breeds with diverse production potential, raised in varied production systems and in several locations (Cobby and Le Du 1978, Papajcsik and Bodero 1988, Dijkstra et al 1997, Landete-Castillejos and Gallego 2000, Cobuci et al 2000, Pollott 2000, Cunha et al 2010, Gloria et al 2012). Another important feature of Wood's model is its simplicity and a reduced number of parameters (Wood 1967)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have