Genetic parameters for traits derived from measures of electrical conductivity collected by automatic milking systems...

Genetic selection for mastitis resistance in dairy cattle has been conducted through both direct selection on clinical mastitis and indirect selection via correlated traits such as SCS. Even with these efforts, mastitis remains a major concern for the dairy industry. Automatic milking systems collect measures of milk electrical conductivity (COND) for individual quarters of the udder at every milking visit, and traits derived from these measures may be good candidates for inclusion in selection indexes as indirect indicators of mastitis resistance. Therefore, the objectives of this study were (1) to derive multiple traits from quarter-level measures of milk COND collected by automatic milking systems; (2) to estimate genetic parameters for traits derived from these measures; and (3) to estimate variance and covariance components between traits derived from measures of milk COND, health traits, and test-day traits in a population of Holstein cattle in the United States. Quarter-level milk COND and visit milk yield were collected at every milking visit on 7,157 Holstein cows milked with automatic milking systems on a farm in north-central Indiana. Additionally, producer-recorded events of clinical ketosis and clinical mastitis, as well as test-day records of fat content, protein content, and SCS, were available for an expanded population of cows that were milked in both automatic milking systems and conventional parlors. Eight traits were derived from quarter-level measures of milk COND collected at each milking visit. These visit-level traits were then summarized into daily values, as either the daily average or the daily maximum, for subsequent analyses. Two additional traits were derived as the coefficient of variation or variance of all quarter-level measures of milk COND for an individual within a day. Additionally, data were analyzed separately by lactation: 1, 2, or 3+. Pedigree data for 53,186 individuals and genotype data for 29,126 individuals, imputed to 50,886 SNP, were used in the analyses. Variance and covariance components were estimated using single-step GBLUP and linear animal models with REML. Estimates of heritability for traits derived from measures of milk COND ranged from 0.01 ± 0.003 (average of COND scores; lactation 3+) to 0.52 ± 0.016 (average of electrical COND; lactation 1). Estimates of genetic correlation between traits derived from measures of milk COND and clinical mastitis ranged from 0.27 ± 0.095 (right-front COND score; lactation 3+) to 0.89 ± 0.054 (coefficient of variation of COND scores; lactation 2), whereas estimates of genetic correlation between traits derived from measures of milk COND and SCS ranged from -0.05 ± 0.095 (quarter extremes ratio of COND scores; lactation 1) to 0.76 ± 0.032 (variance of COND scores; lactation 3+). Our results show that traits derived from measures of milk COND are heritable, and they have moderate to high genetic correlations with clinical mastitis, as well as near-zero to high genetic correlations with SCS. These results suggest that traits derived from measures of milk COND are promising indicators of clinical mastitis, and incorporating them into selection indexes in complement to traits already included in indexes for udder health can enhance genetic selection for mastitis resistance.