NEW APPROACH OF PREDICTING MILK PRODUCTION RESPONSES AND FEED EFFICIENCY TO CHANGES IN DIET COMPOSITION TO BE USED IN NORFOR ONFARM SOFTWARE

Feeding dairy cattle is one of the bases for a good milk production. Of the total costs feed represents up to 80%. Moreover, price of milk and feed are very volatile, making the calculation of margins very difficult to predict and forecast. Thus, it is vital to feed the cows with precision, with few leftovers, and assuring that the majority of the feed they consumed is transformed into milk, this is called high feed efficiency. Together with this, precise and accurate feeding of cows will help to reduce the environmental impact. In the Nordic countries, the most popular system used for cow nutrition is the NorFor System (Nordic Feed Evaluation System). The goal of the NorFor System is to predict the true nutritious value of the ration, an accurate predict feed intake and the cows digestion and metabolism of the given diet. With all this, the final goal is to formulate rations for the cow that are precise and have an economic optimum with the least environmental impact possible. This system is based on equations that try to predict all the stated above. This project main objective is to keep improving the NorFor system to keep up with the challenges of modern dairy production by including new approaches and tools. It will focus in two main aspects of the NorFor system: Firstly, the evaluation of the feed nutritious value, by trying to create a method to estimate accurately the energy content of concentrate mixes (feed products that have more than one ingredient). Secondly, the cow's responses to changes in the diet. The project will try to change the approach on how to feed the cow based on their responses to changes in the diet composition rather than a static feeding system based on the current production level. In this matter, we will also try to include Machine Learning analysis in the data collected from farms to adjust our findings to each farm reality. Results showed that empirical models including organic matter digestibility (OMD) and chemical components can be used to estimate energy contents of compound feeds when information to use the mechanistic model is not available. Regarding the response functions, this project was able to develop a model, based on previous studies, of milk responses to changes in silage digestibility and changes in the concentrate intake. In addition, first attempts to adapt this model to real-time farm data. The adaptation of this model into farms will allow farmers to take decisions about feed efficiency and profit.

Based on the developed model to estimate energy content of compound feeds, NorFor will include this model in their system and use it as a practical methodology. Responses of milk to diet changes and its adaptation to real-farm time data will allow farmers and advisors to asjust concentrate levels according to the fed forage, and improve feed efficienct and profit per kg of milk.

Feed represent one of the major costs in modern dairy production. The real feed value of the diet, the animal products, and the partial efficiency of feed utilization for the particular livestock product, all influence total efficiency, and thus overall feed costs. Continues development of our common Nordic feed evaluation system, NorFor, is an important step towards implementing more sophisticated nutritional strategies to control production responses from dairy cows and improve feed utilization. In a context of a higher volatility of feed and milk prices, quantification of animals multiple responses to dietary changes is of particular interest to help dairy farmers in optimizing the feeding and reducing the environmental impact. The main objective of this project is to improve the NorFor system by developing a dynamic dairy cow nutrition approach. This means moving from a diet formulation based on a pre-defined production level to a response approach, adjusting the diet based on the response to changes in diet composition. This should improve feed efficiency and improve financial return by a more precise feeding. Improved feed efficiency will also reduce greenhouse gas emissions. This new approach will be implemented to the NorFor software system. Farm variation has been one of the biggest challenges for adaprtation from R&D into the commercial area. Thus, a big challenge for this project is the intention to bring the "Information and Communication Technologies" (ICTs) into the Animal Science Field, like Machine Learning (ML). With ML, the goal is to bring the data collected from farms by sensing technologies (body changes, milk prodcution changes, etc.) for accounting farm variation. Other big Challenge is validation of the models. The Project will test the dynamic modelling approach on farms in Denmark, Norway and Sweden by implementing the models into national NorFor Client software Tools.