CONICET | Buscador de Institutos y Recursos Humanos

The objective of the present study was to assess therelationship between individual cow milk yield andfertility, accounting for the contextual effect of theherd. A data set including 657,968 lactations from 677dairy herds in Argentina from 2001 to 2012 was used.The odds of pregnancy by 100 d in milk (DIM) wereassessed by a multilevel logistic model (with cow asthe first and herd as the second hierarchical level),and time to pregnancy was assessed by a proportionalhazards regression model. Multilevel logistic models includedthe fixed effects of milk yield by 80 DIM, parity,year, and calving season at cow level and quartiles ofherd milk yield by 80 DIM as a contextual effect. Theproportional hazards model included the effect of dailycow-level milk yield as time-dependent variable, withmilk yield at herd level as the stratification variable.Cows producing 1 standard deviation over the meanmilk yield of their herd had 1.3 percentage point lowerpregnancy by 100 DIM (from 31.4 to 30.1%; odds ratio= 0.942) when in herds in the top quartile of milk yield,whereas they increased 0.5 percentage points (from 27.9to 28.4%) when in herds in the lowest quartile of milkyield. Only 4% of the observed variation in pregnancyby 100 DIM was explained by the random effect of theherd. Similarly, cows producing 1 standard deviation (8kg/d) greater than the herd mean daily milk had 1.3%lower hazard of pregnancy (hazard ratio = 0.987) at 63DIM in herds in the top quartile of milk yield, whereasthey had 14.8% higher hazard (hazard ratio = 1.148)in herds in the lowest quartile of milk yield. The magnitudeof the negative association between the cow?sdaily milk yield and the hazard of pregnancy increasedwith DIM. In conclusion, the relationship between milkyield and reproductive performance is statistically significant,but the effect size is practically small and ismodulated by herd production level.