Exploratory analysis of associations between postpartum body condition changes measured by an automated 3-dimensional camera and reproductive outcomes measured by in-line milk progesterone analysis.

Our objective was to evaluate the associations of postpartum BCS changes, as predicted by automatic monitoring using 3-dimensional (3D) camera technology after calving, with milk yield and reproductive outcomes assessed both from in-line milk progesterone analysis and conventional methods in Holstein dairy cows in a retrospective observational study. Cows calving in a commercial farm during a whole year (n = 123; 53 primiparous and 70 multiparous) were automatically evaluated for BCS with a 3D camera (DeLaval Body Condition Scoring system) daily on a 5-point scale up to 120 DIM and retrospectively classified into tertiles based on BCS change from calving to 30 DIM. The resulting groups had high (HI), intermediate (IN), and low (LO) BCS change. Milk yield and milking frequency were collected daily from the automatic milking system (VMS V310, DeLaval), which also automatically evaluated milk progesterone concentration every 2 to 3 d on average. These data were used to characterize luteal activity, with progesterone concentrations peaking during luteal phases and reaching a nadir around estrus events, thus allowing identification of the resumption of cyclicity. Pregnancy outcomes were evaluated up to 200 DIM with logistic regressions and time-to-event data with Cox's proportional hazard models. Continuous data were analyzed with ANOVA, with repeated measures mixed models when appropriate. The models included the effects of BCS tertile and parity. Body condition score loss averaged -0.47, -0.30, and -0.09 for HI, IN, and LO, respectively, ranging from -0.78 to -0.36 in HI, from -0.35 to -0.25 in IN, and from -0.24 to 0.41 in LO. Cows with minimal BCS loss (LO) were thinner (BCS = 3.23 ± 0.29) at calving than cows with greater BCS loss (HI and IN, BCS = 3.38 ± 0.25 and 3.37 ± 0.23, respectively). The BCS nadir was lower in HI compared with IN and LO, although the time to the nadir did not differ. The HI cows tended to have later commencement of luteal activity (i.e., DIM of the first of at least 2 consecutive samples with milk progesterone ≥5 ng/mL; +8 d vs. IN and LO), calculated based on progesterone profiles, but a similar percentage of cows (∼95% overall) resumed ovarian cyclicity before the end of voluntary waiting period (75 DIM). The proportion of cows pregnant at the first artificial insemination (AI) did not differ, but LO cows tended to have increased likelihood of pregnancy at the second AI compared with IN. The cumulative proportion of pregnant cows at the first and second AI tended to be greater in LO compared with HI and IN, and the proportion of pregnant cows up to 200 DIM was greater in LO. Compared with LO, the HI and IN cows had reduced hazard of pregnancy up to 200 DIM. Milk yield was 4 and 3 kg/d lower in LO compared with HI and IN. Automated BCS and progesterone monitoring showed that greater BCS loss in the first 28 DIM was associated with delayed cyclicity and reduced reproductive performance, whereas minimal loss was associated with greater reproductive success but lower yield.