Phenotypic and genetic characterization of novel somatic cell count traits from weekly or monthly observations

The objectives of this study were (1) to explore traits that better capture weekly or monthly changes in somatic cell counts (SCC) than does the commonly used lactation-average SCC, (2) to estimate their heritabilities and relationships to clinical mastitis (CM), and (3) to determine if these traits are feasible for use in monthly testing schemes. Clinical mastitis and weekly test-day (TD) records of SCC and milk production traits from 1,006 lactations of Swedish Red and Holstein cows collected from 1989 to 2004 were used (data set W). A data subset was also created to mimic monthly recording (data set M, 980 lactations). Twenty SCC traits were defined, taking into account SCC general levels and variation along the lactation curve, time and level of infection, and time of recovery. To reduce dimensionality, cluster and stepwise logistic regression procedures were applied. In data set W, 3 traits, "standard deviation of SCC over the lactation," a discrete (0/1) indicator of "at least one TD with SCC >500,000 cells/mL", and "number of days sick in the widest SCC peak" (DWidest) were the variables kept both with cluster procedures and a stepwise logistic regression with the logit of CM as dependent variable. In data set M, DWidest was replaced by "number of SCC peaks" and "average number of days sick per peak" (ADSick). Lactation-average SCC (in the first 150 d or between 150 and 305 d) did not enter into the logistic regression. Heritability estimates obtained for these new traits under a Bayesian setting and a Gibbs sampling approach were 10 to 16% (except for ADSick: 5%). Heritabilities were at least as high in the monthly data set as in the weekly data set. Thus, these SCC traits seem promising for use in breeding programs based on monthly milk recording.