Accounting for voluntary waiting period changes in U.S. dairy herds: Adjusting daughter pregnancy rate and introducing first-service to conception

Authors: GRAHAM, JASON - Holstein Association Usa, Inc; MCWHORTER, TAYLOR - Council On Dairy Cattle Breeding; COLE, JOHN - Council On Dairy Cattle Breeding; DURR, JOAO - Council On Dairy Cattle Breeding; NICOLAZZI, EZEQUIEL - Council On Dairy Cattle Breeding; LEGARRA, ANDRES - Council On Dairy Cattle Breeding; MILES, ASHA - Dairy Records Management Systems(DRMS); Hutchison, Jana; VANRADEN, PAUL - Retired ARS Employee (USA)

Submitted to: Journal of Dairy Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/27/2026
Publication Date: N/A
Citation: N/A

Interpretive Summary: Dairy farmers rely on cows becoming pregnant on time to maintain steady milk production and keep herds profitable. However, modern dairy cows often produce milk for longer periods, causing many farmers to delay breeding after calving. Current national fertility evaluation tools assume all farms breed cows at the same time after calving, an assumption that no longer reflects today’s management practices. As a result, fertility rankings used for selecting bulls and cows may not match real world conditions on farms. Researchers analyzed more than 32 million records from U.S. dairy cows to update how fertility is measured. A new trait, called First Service to Conception, was evaluated to capture how long it truly takes a cow to become pregnant. An improved version of an existing fertility measure was also developed to adjust for each farm’s actual breeding schedule instead of relying on a fixed timeline. These improvements provide a more accurate picture of reproductive performance and lead to more meaningful rankings of cows and bulls used in breeding programs. More accurate fertility evaluations help farmers select animals that become pregnant more consistently, reducing costly delays in reproduction. Because the updated methods change the ranking of top bulls, as shown by correlations as low as 0.88, these improvements influence breeding decisions across millions of inseminations nationwide. When selection tools align with modern farm practices, farmers benefit from fewer failed breedings, more timely pregnancies, and more efficient milk production cycles. At the national level, even modest gains in fertility accuracy can save the dairy industry many millions of dollars each year through improved reproductive efficiency and faster genetic progress. This advancement strengthens the productivity, profitability, and long term sustainability of the U.S. dairy sector.

Technical Abstract: Integrating functional annotations into genome-to-phenome studies remains challenging in farm animals. Stratified LD score regression relies on approximations for human populations and fails under strong linkage disequilibrium. Genomic feature BLUP models typically accommodate only two annotation categories at a time, failing to account for confounding effects among multiple annotations. Furthermore, we found that large-effect QTLs distort heritability partitioning in multi-component genomic REML models. Their effects can be incorrectly absorbed by small functional categories, severely inflating variance estimates for these categories. We developed a unified framework with two complementary approaches. GEMRICH, our newly developed R package, estimates enrichment of causal variants in functional annotations based on statistical fine-mapping of significant association regions. MPH, a fast REML implementation, enables simultaneous heritability partitioning across many functional annotations while modeling fine-mapped lead variants as fixed effects, preventing large-effect QTLs from inflating polygenic variance estimates. This quantifies genetic architecture comprehensively: causal variant enrichment in significant regions and per-SNP heritability enrichment of sub-threshold polygenic effects. Simulation analyses confirmed this strategy eliminates bias in heritability partitioning caused by large-effect QTLs. We analyzed 38 dairy cattle traits using ~50,000 Holstein bulls with 11 million imputed sequence variants. Analysis across genic regions, evolutionarily constrained elements, cis-eQTLs and cis-sQTLs, chromatin states, and ATAC-seq peaks from over 40 tissues revealed substantial heritability enrichment (e.g., 5–10-fold in coding sequences, ~10-fold in constrained elements, and 2–5-fold in regulatory regions). Tissue-specific analyses revealed particularly strong enrichment in mammary tissue annotations for yield traits, and bone marrow annotations showed broad relevance across diverse dairy traits. Incorporating functional enrichments into genomic prediction improved accuracy over chip-SNP-based GBLUP by 1.17% on average across 36 traits (two traits excluded due to insufficient validation bulls), with gains of up to 3.42% for cow livability, using old–young validation with December 2020 training and December 2021 validation data. To enable broad adoption, we released open-source software tools at github.com/jiang18/gemrich and jiang18.github.io/mph.