Development of innovative and practical management tools to improve sustainability of milk production and quality of dairy products

In the current complex economical context, novel strategies are needed to help local dairy farmers to face the European dairy sector crisis. This thesis was initiated in the framework of ManageMilk project and was globally aimed to investigate the possibility to develop some innovative and practical management tools helping dairy farmers in their daily decisions. To develop such management tools, several conditions must be fulfilled. Firstly, used data must be relevant. According to the literature, the milk composition, and in particular, the milk fatty acid (FA) profile, appears to be a suitable trait allowing useful information about the dairy cow’s health status or about the management system efficiency. These data must also be easily available at low cost from milk recording organization. Recently, the MIR spectrometry offers the possibility to build routinely cheaper and more important databases. To develop management tools, milk samples have to be collected using comparable sampling methods. Unfortunately, in order to decrease the milk quality control costs, the International Committee for Animal Recording allows alternative sampling schemes including the collection of samples from morning or evening only milkings. This alternative sampling scheme can interact with phenotypic and genetic parameters. Therefore, additionally to the development of conversion equations, this thesis is establishing if morning or evening only milkings are genetically different traits. Last condition concerns a useful phenotypic and genetic variability. Milk FA profile is, among others, altered by genetics. So, one paper of this thesis concerns the setup of a useful genetic evaluation model able to estimate accurately the genetic part of milk fat composition variations. Routine genetic evaluation of production traits in dairy cattle commonly uses random regression model (RRM). Recently, “splines” have been advocated as a good alternative to Legendre polynomials (LP) for analyzing test-day yields in RRM. Therefore, several models are compared. Obtained results show the possibility to propose a practical and robust method for estimating accurate daily major FA production from single milking, useful for a further development of practical management tools helping dairy farmers in their daily decisions.