Submitted to: BARC Poster Day
Publication Type: Abstract Only
Publication Acceptance Date: 5/1/2003
Publication Date: N/A
Technical Abstract: In dairy cattle, female fertility is correlated antagonistically with production traits. Selection for increased yield, without concomitant selection for fertility, has resulted in a genetic and phenotypic decline in female reproductive ability over the last 30 to 40 years. Days open (DO), defined as number of days from calving to successful breeding, is a common measure of fertility in dairy cattle, with larger DO indicating poorer fertility. Since 1965, DO has increased about 40 days in U.S Holsteins. To limit this decline, AIPL implemented a new, national genetic evaluation for female fertility in February, 2003. The evaluations are reported as daughter pregnancy rate which is calculated from DO as (233 - DO)/4. To allow for selection to occur as soon as possible, the ideal is to use records in genetic evaluation as soon as they are available. However, use of early, incomplete DO records could introduce bias in genetic evaluations since only the more fertile cows would have a record early in lactation. If a sires' daughters have truly poor fertility, for example, only his best daughters would have a DO record early in lactation and an evaluation based on only those most fertile daughters would give a biased evaluation for that sire. To avoid this potential bias, cows are currently required to have a minimum of 250 days in milk to be included for genetic evaluation. Furthermore, DO is set to 250 for records that exceed the upper limit. The objective of this research was to develop a procedure to predict DO for "incomplete" records. This would allow use of DO early in lactation and avoid the potential biases previously described. Additionally, reliance on a 250 day upper limit, to avoid extreme observations, could be eliminated by use of a projected record instead. Solutions from a simple, fixed, linear model were used for projection for each of 10 DO groups. DO groups were defined in 20 day increments starting with 70 and up to 250. Along with an overall intercept, the model for prediction, in each DO group, included the effects of lactation, age, milk yield, calving ease score, days open based on most recent breeding ("DOx", x=70,90, ....,250), and regression on previous DO and previous number of services for second and later lactations. Quadratic effects were also included for the covariates of age and milk yield. Phenotypic correlations between predicted and actual DO ranged from .41 (DO group 70) to .87 (DO group 250). Mean prediction error was basically 0 in each group. Standard deviation of prediction errors ranged from 27 to 54 days. These results clearly indicate the potential to use DO records prior to 250 days in milk and perhaps as early as 90 days. Further research will focus on development of weighting factors for projected records and the effect of projected records on sire evaluations.