Dry matter intake and efficiency in lactating Holstein cows grouped by direct genomic values for feed utilization

Abstract Text: Feed costs represent the largest expense for lactating dairy cows, yet the dairy industry does not currently select directly for feed efficiency.  The purpose of this study was to observe how accurately genomic predictions of feed utilization developed from a small reference population could predict feed utilization in lactating Holsteins.  Feed intake data was collected from 970 Holstein cows across 11 commercial Pennsylvania herds.  Genetic evaluations of four feed utilization traits were conducted: dry matter intake (DMI), dry matter efficiency (DME; energy corrected milk / DMI), crude protein efficiency (CPE; protein yield / protein intake), and residual feed intake (RFI).  A genome-wide association study was performed using 45,138 single nucleotide polymorphism (SNP) genotypes from 340 cows with the e-BIGS program (http://bigs.ansci.iastate.edu), and SNP associations were used to generate direct genomic values (DGV) of feed utilization.  Cows were subsequently separated into feed utilization groups with pen milk yield and feed intake monitored for two weeks.  High and low DGV groups of twenty cows for each of the four feed utilization traits were observed at Penn State, whereas cows from a commercial dairy farm were divided into high, medium, and low DGV groups of 51 cows for DMI and 68 cows for DME.  Mixed models were used to evaluate differences in feed utilization between high, medium (commercial herd only), and low DGV groups and included DGV group, a random effect for date, and random error.  The high DMI group consumed 10.8 kg more dry matter per day (P<0.01) than the low DMI group at Penn State.  Likewise, the high CPE group produced 0.04 kg more protein per kg of protein intake (P<0.05). The high DME group tended (P<0.10) to produce more milk (+0.17 kg) per kg of intake than the low group.  Differences among high and low RFI groups were opposite of expectations, with the low group consuming significantly more dry matter than predicted (P<0.01) based on milk yield and body weight compared with the high group.  Differences in DMI (p=0.36) and DME (p=0.14) were not significant among groups on the commercial dairy farm.  This research demonstrated that it is feasible to apply genomic predictions of feed utilization developed from a reference population to select for improved efficiency in commercial dairy herds.  However, not all the predictions were as expected and a larger reference population is required to increase prediction accuracy.

Keywords: feed efficiency, genomics