Lifetime feed efficiency and deep phenotypes from scarce feed intake records using the mechanistic LiGAPS-Dairy model

Ideally, selection for feed efficiency requires deep phenotyping of net efficiency, or lifetime recording of intake and all energy sinks across environments. However, recording of feed intake is scarce. Therefore, net efficiency is often defined as a simplistic linear equation, e.g. RFI. We tested the use of the mechanistic LiGAPS-Dairy model to derive nine deep phenotypes with a dataset for 1,228 dairy cows, combining feed intake, yield and liveweight data, with ration, weather, cow and farm data. Mismatch between data recording and model assumptions made this process time consuming, but allowing for missing parities and further automation should improve this quickly. We managed for 206 cows to estimate the deep phenotypes. Heritability and phenotypic correlations between the nine traits were estimated. When the pipeline is finished, the mechanistic LiGAPS-Dairy model will enable us to derive a more comprehensive breeding goal, more closely resembling net efficiency, whilst utilising scarce records.