Phenotype-Based Prediction of Selection Indices in Dairy Cattle without Pedigree or Genomic Information

Abstract New Zealand dairy farmers rely on industry-supported genetic indices in the National Breeding Objective, including Breeding Worth, Production Worth, and Lactation Worth to guide selection and culling. Animals without pedigree or genomic data receive low-reliability breed-average values, limiting early decisions. We developed phenotype-based “New Zealand Entry Evaluations” (NZEE) that predict BW, PW and LW from early-lactation records. First-lactation heifers (n = 60; Friesian, Jersey, or dual-purpose [Norwegian Red/Fleckvieh]) in a spring-calving pasture herd were recorded at ~100 days in milk for milk volume, fat, protein and somatic cell count; mature liveweight breeding value (BVW) was used as a body-size proxy. Multiple regression models predicted NZEEBW, NZEEPW, NZEELW, FATBV and PROTBV without pedigree or genomic inputs; breed-group means were compared by one-way ANOVA. NZEE predictions correlated strongly with industry indices (r = 0.87 for LW, 0.82 for BW, 0.75 for PW), explaining 55–78% of variation. NZEEBW aligned more closely with measured milk solids than pedigree-based BW. Breed effects were significant for fat and protein breeding values (Friesian vs Jersey; P < 0.05) but not for NZEEBW, NZEEPW or NZEELW (P > 0.05). These models enable earlier ranking of non-pedigreed heifers using routinely collected on-farm data. Information © The Authors 2026