The animal Article of the Month for January is entitled ‘Genomic selection for feed efficiency in dairy cattle’

Feed is a major cost variable in livestock production systems, so dairy farmers would like to breed feed-efficient cows. However, feed efficiency is rarely part of the breeding objective in cattle (unlike, say, pigs and poultry). Genomic selection offers a way to predict the feed efficiency of individual cows but the accuracy of predictions requires a much larger pool of data than currently exists. Sharing genomic data between countries could provide the way forward.

Genomic selection allows a trait associated with feed efficiency (such as dry matter intake) to be incorporated into a multi-trait breeding objective that includes all traits that contribute to profitability. This is currently the way many countries are seeking to improve feed efficiency. There are other ways to calculate feed efficiency, primarily ratio traits or residual traits, but each has its problems. The ratio of the amount a cow eats in a period compared with its product (milk or meat) requires large amounts of intake data measured on expensive, labour-intensive equipment. Because of the cost, this type of measurement in beef or dairy cattle is generally limited to research herds. The other option is residual (or net) feed intake (RFI): the feed that is left over after accounting for the feed required to maintain the cow and allow for her level of production. The advantage of RFI is that it captures variation in activity, protein turnover, heat increment and digestibility. However, RFI in lactating cows is complicated: in early lactation, cows tend to mobilise their own body tissue to sustain lactation, and have poor fertility at this time.  Ideally, RFI should be adjusted for changes in body reserves, but they are difficult to accurately quantify. For the time being, multi-trait selection indices appear the best way to select for feed efficiency.

To incorporate feed efficiency into a multi-trait selection index, we need breeding values. Breeding values are produced by estimating associations between many thousands of genetic markers such as SNPs (single nucleotide polymorphisms) and the feed efficiency phenotype to produce a genomic prediction equation that can be applied to cows or bulls that have the SNP data. This process gives all selection candidates genomically predicted breeding values of feed efficiency that can be used for selection.

The accuracy of the resulting predictions depends on the volume of available data. For traits where a lot of data is collected, such as milk yield, accuracies of genomic prediction of 0.8 are currently achievable. However, the accuracy of feed efficiency traits is only around 0.4; not enough to give farmers confidence in the technology. Much larger reference populations need to be assembled to improve this accuracy. Although multi-country reference populations present issues such as variability in measurement techniques, diets fed and even climate and location, they have been tested with some success. Genomics with multi-country collaboration could revolutionise the way we select for feed efficiency in cattle, by using data collected on a small scale to generate genomic breeding values that enable genomic selection in an entire population.

Access the full paper here

Authors: J. E. Pryce, W. J. Wales, Y. de Haas, R. F. Veerkamp and B. J. Hayes

The animal Article of the Month is selected by the Editor-in-Chief and is freely available for one month