Towards an improved estimation of animal feed efficiency

Improving feed efficiency is essential for sustainable livestock farming: it is expected to reduce feed resources use and to decrease waste and environmental impacts. Ruminants are key players because they transform resources, which are non-edible for human consumption, such as grass, into edible resources for human consumption, such as milk or meat. Improving the use of land and feeds by animals is important because arable land availability decreases and because feed accounts for the largest part of production cost.

Improving feed efficiency can be done by selecting the most efficient animals. This selection requires an estimation of feed efficiency variability to estimate the genetic progress, which is potentially reachable. Feed efficiency estimation requires measurements of individual animal feed consumption and what we expect the animal to do with this feed. For example, a dairy cow will usually use the feed to maintain her vital functions, to produce milk in volume and quality, but also to gain body reserves, which are an essential resource after calving. To determine feed efficiency, a cow is always compared to other cows sharing the same environment. A cow is efficient if she produces as much as the average of the group of cows, but eats less feed than the group average (or eats the average but produces more). The difference between the actual feed consumed and the feed that she should have consumed given her productive outputs is called residual feed consumption. This residual feed consumption encompasses differences in efficiency but also errors, such as measurement errors or errors in the calculation of expected feed consumption from measured outputs, which are not interpretable as feed efficiency - even if this is often done in practice.

The current paper tackles this issue in dairy cows by proposing a method to isolate the feed efficiency part of residual feed consumption from the error part. The main interest in livestock farming is to have cows, which are repeatably efficient over time. The feed efficiency part of residual feed consumption was therefore defined as the repeatable part of residual feed consumption specific to cows, and the error part as the random non-repeatable part of residual feed consumption.

A usual interpretation of residual feed consumption would conclude that in our study, feed efficiency differences account for 8% of the actual variation in feed consumption. In fact, our study shows that only 4.7% (59% of the 8%) of the measured feed consumption variation was repeatable and specific to cows, therefore interpretable as feed efficiency. When the goal is to estimate the variability of feed efficiency using the residual feed consumption method, attention has to be paid to its interpretation because only 59% of the observed variability might really be associated with feed efficiency and not with errors.