The aim of this study was to gain a better understanding of the variability in shape and features of all progesterone profiles during oestrus cycles in cows, and to create templates for cycle shapes and features as a base for further research. Milk progesterone data from 1418 oestrus cycles, coming from 1009 lactations, was obtained from the Danish Cattle Research Centre in Foulum, Denmark. Milk samples were analyzed daily using a Ridgeway ELISA-kit. Oestrus cycles with less than ten data points or shorter than four days were discarded, after which 1006 cycles remained in the analysis. A median kernel of three data points was used to smooth the progesterone time series. The time between start of progesterone rise and end of progesterone decline was identified by fitting a simple model consisting of base length and a quadratic curve to progesterone data, and this luteal like phase (LLP) was used for further analysis. The dataset of 1006 LLP’s was divided into five quantiles based on length. Within quantiles, a cluster analysis was performed based on shape distance. Height, upward and downward slope, and progesterone level on day 5 were compared between quantiles. Also, the ratio of typical versus atypical shapes was described, using a reference curve based on data in Q1-Q4. The main results of this paper were that 1) the majority of the progesterone profiles showed a typical profile, including the ones that exceeded the optimum cycle length of 24 days; 2) cycles in Q2 and Q3 had steeper slopes and higher peak progesterone levels than cycles in Q1 and Q4 but, when normalized, had a similar shape. Results were used to define differences between quantiles that can be used as templates. Compared to Q1, LLP’s in Q2 had a shape that is 1.068 times steeper and 1.048 times higher. LLP’s in Q3 were 1.053 times steeper and 1.018 times higher. LLP’s in Q4 were 0.977 times steeper and 0.973 times higher than LLP’s in Q1. This paper adds to our knowledge about the variability of progesterone profiles and their shape differences. The profile clustering procedure described in this paper can be used as a means to classify progesterone profiles without recourse to an a priori set of rules, which arbitrarily segment the natural variability in these profiles. Using data-derived profile shapes may allow a more accurate assessment of the effects of for example nutritional management or breeding system on progesterone profiles.
Theriogenology, 2016, Vol 86, Issue 4, p. 1061-1071