Dealing specifically with the microbiota x animal interaction, growing evidence indicates that rumen microbiota diversity may be affected by animal factors as physiological status (transition period), milk yield, animal parity, breed, and feed efficiency (Bickhart and Weimer, 2018 ). Moreover, ruminal diversity of microbiota may be different among individuals, and it could exhibit properties of resilience by returning to preliminary microbial composition after a dietary perturbation (Weimer et al., 2010 ). This indicates that there is a direct causal relationship between animal phenotype and a specialized rumen microbiota. Currently, there is a need to phenotype individually rumen microbiota to characterize its diversity among animals, in order to select animals with the most appropriate ruminal diversity susceptible to adapt rapidly to changes in feed resources.
Proposed research questions and associated strategy. In the continuation of the previous MoSAR project, we propose to phenotype ruminal process in animals fed different extreme diets (e.g., acidotic diets vs high-fiber diets, low- vs high-lipid diets to mitigate CH4 emissions, etc..) in order to characterize 

the adaptive capacity of the microbial population (changes in diversity and activities) and its relations to digestive efficiency. This approach will be combined with the animal specificity in term of personality and ingestive behavior (cf. interface diet x animal), to explore the association between feeding behavior, changes in microbial population composition and digestive efficiency at different scales of time along their lifespan. This will help to predict the respective parts of microbiote function and animal host to adaptive capacity over time in order to quantify the best combinations of microbiote x animal for better feed efficiency and reduced wastes. These approaches will combine in vitro and in vivo approaches to quantify rumen metabolism associated with the development of metagenomics methods to characterize the diversity of ruminal populations, that was initiated during the previous five-year period.

Associated to experimental works, on the basis of our work on dynamic modelling (Muñoz-Tamayo et al., 2016), modelling works will go on structuring a mechanistic generic model of rumen metabolism and function from the existing models that already predict pH, VFA production, microbial synthesis, and ruminal biohydrogenation of fatty acids (FA) on a daily basis. Attempts will be made to integrate more closely the enzyme kinetics, the thermodynamic driving forces of rumen, and the activities of the microbiota to have a better understanding of rumen function and possibly to control the rumen driving forces in order to help to improve feed evaluation and efficiency. One of main objectives will be to integrate individual animal variability for digestive robustness and efficiency into that generic model. In parallel, it will contribute to elaborate tools to complete actual feed unit system (Sauvant and Nozière, 2016) with the objective to integrate the digestive responses, in particular the metabolism of FA to further predict the polyunsaturated FA need of lactating females. 

Given the advantages of omics technologies to characterize the rumen ecosystem, a flagship of MoSAR research will be the investigation of how microbial omics information can be integrated into mathematical models that can inform on how to engineer the rumen microbiome. These developments will be supported by existing strong collaborations with national and international partners in the domains of microbiology, omics and computer science.