Pregnancy per AI was fitted as a binary response variable with a logit link function to a
129 generalized linear mixed model for repeated measures by using logistic regression (GLIMMIX
130 procedure of SAS; SAS Inst., Inc., Cary, NC). If cows did not become pregnant they shifted to a
131 later cohort and may have been on the same or the alternate synchronization protocol depending
132 on the date on which they were available for resynchronization. Since cows had the opportunity
133 to be measured more than once and could be on either treatment, a repeated measures analysis
134 was conducted. The final list of independent variables fitted in the model were treatment, month
135 of breeding, insemination number (first insemination or resynchronization), and lactation number
136 (1, 2 or 3 or more) fitted as categorical variables, and SCC, estimated milk production for the
137 current lactation (305ME), milk fat percentage, milk protein percentage, and milk urea nitrogen