Genomic signatures of long-term selection in the Langhill dairy herd: the power of a five-decade experiment for sustainable dairy cattle breeding

Background Selective breeding has transformed dairy cattle productivity, but the genomic consequences of long-term directional selection - what it costs in diversity and where in the genome it acts - are hard to quantify because contemporaneous control populations are rare. The Langhill experiment, established in 1973, is one of the world's longest-running dairy cattle selection trials with a designed control line kept under common management. We used this resource population to ask what five decades of bull selection have cost in genetic diversity, where in the genome selection has acted, and whether complementary scans converge on the loci under selection. Results We genotyped archived blood samples from 1,126 cows of two genetic lines: a Select line sired by bulls in the top decile of UK genetic merit for milk fat-plus-protein yield, and a Control line sired by bulls of average merit. The Select line showed the expected gains in every yield trait. Inbreeding accumulated at the same annual rate in both lines, and effective population size, estimated from that rate, was approximately 33. Linkage disequilibrium was higher in the Select line at every marker distance from 2 to 300 kb. Autozygosity was 24.7% higher in the Select line than in the Control line (0.078 versus 0.063), a difference robust across four calling parameter sets and two independent callers. The excess was confined to runs of homozygosity longer than 2 Mb; shorter runs did not differ between lines, confirming a shared founder base and placing all divergence after the lines were split. It did not arise from a numerically narrower sire pool: the Select line used more sires (79 versus 72), each with fewer daughters, yet its cows were three times more related to one another. Twenty-eight regions across thirteen chromosomes showed strong differentiation; intersecting five scans identified a four-method signal on chromosome 21 spanning the SERPINA gene cluster and sixteen three-method regions, including chromosome 3, with haplotype and diversity statistics agreeing on the line carrying the signal in fourteen of seventeen. Autozygosity itself did not localise to any region. Conclusions Selection left two genomic footprints: diffuse, recent inbreeding accumulated through the narrow ancestry of the elite sire pool, and localised differentiation at specific loci. Runs of homozygosity index only the first process; the second is captured by frequency- and haplotype-based differentiation scans. Because the cost of inbreeding is incurred through ancestry rather than the number of bulls used, mating schemes that constrain the rate of co-ancestry would be preferable to simply managing the number of sires selected and used.