Genetic analyses of udder conformation traits and daily milk yield measured by robotic milking systems using repeatability and random regression models in American Holstein cattle.

Udder conformation is a key component of functional dairy production because it influences milking efficiency, susceptibility to mastitis, and cow welfare and longevity. The increasing adoption of automatic milking systems (AMS) enables repeated, objective measurement of teat and udder geometry at each milking, thereby creating new opportunities to improve genetic evaluation of these traits under commercial conditions. In this study, we estimated genetic parameters for 5 AMS-derived udder conformation traits and daily milk yield (DMY) in American Holstein cows using repeatability and random regression models (RRM). A total of 10,422,361 milking events from 7,546 cows (after quality control) were recorded by 36 AMS units on a large commercial farm in Indiana and aggregated into 3,804,166 daily records for udder depth (UD), front teat distance (FTD), rear teat distance (RTD), distance front to rear teats (DFR), udder balance (UB), and DMY. Cows were genotyped for 60,499 SNPs. Repeatability models were fitted within and across lactations, whereas RRM were fitted separately by lactation using Legendre orthogonal polynomials (up to fifth order), considering homogeneous or heterogeneous residual variance structures. In both modeling approaches, (co)variance components were estimated via the average information REML algorithm under a GBLUP framework using the BLUPF90 software. Using the repeatability model across all lactations, h2 estimates (±SE) ranged from 0.17 ± 0.01 (DMY) to 0.69 ± 0.01 (UD), and repeatability was high for all udder conformation traits (0.88 ± 0.01 to 0.95 ± 0.01) and moderate for DMY (0.59 ± 0.01). Using the repeatability model within lactation, h2 estimates ranged from 0.21 ± 0.02 (DMY) to 0.79 ± 0.02 (UD), with high repeatability for udder conformation traits (0.86 ± 0.01 to 0.96 ± 0.01). The analyses based on RRM further revealed DIM-specific h2 estimates ranges of 0.33 to 0.71 (UD), 0.36 to 0.63 (FTD), 0.10 to 0.47 (RTD), 0.40 to 0.64 (DFR), 0.15 to 0.44 (UB), and 0.13 to 0.33 (DMY). The genetic correlation estimates for the same udder conformation trait across lactations were consistently high (>0.80), indicating substantial genetic stability across parities. Within lactation, genetic correlations across DIM were also high (>0.70), particularly in later parities. Across all lactations, FTD and RTD were moderately and positively genetically correlated, whereas UD exhibited negative genetic correlations with teat spacing and geometric traits (e.g., FTD, RTD, and DFR). Genetic correlations between udder conformation traits and DMY were weak to moderate (up to 0.37 under the repeatability model). Collectively, these results indicate that AMS-derived udder conformation traits are under moderate to high genetic control, which is consistent across DIM and lactations, supporting their incorporation into genomic selection programs in dairy cattle. The RRM provided additional resolution on within-lactation dynamics, whereas repeatability models captured most of the additive genetic signal required for routine genetic evaluations.