Possibility to Estimate Same Day Energy Status of Dairy Cows during First Half of Lactation by Non-Invasive Markers with Emphasis to Milk Fatty Acids

Abstract

Postpartum negative energy balance (NEB) is detrimental to cows and decreases profitability in dairy farming. The two origins of milk fatty acids (FA), de novo synthesized in the mammary gland and plasma lipids initially originating from feed, rumen microbes and the animal's adipose tissue, make milk FA candidates as possible NEB biomarkers. The aim of this study was to assess the possibility to predict EB in cows in the first 150 days of lactation with BCS, milk traits and selected individual milk FA and the ratios of blood-derived and de novo synthesized FA. The daily EB of Estonian Holstein cows (N = 30) was calculated based on body weights and BCS values. Milk FA were analyzed with gas chromatography. The variance partitioning analysis revealed that milk production traits, BCS at calving, FA ratios and days in milk accounted for 67.1% of the EB variance. Random forest analysis indicated the highest impact of the ratios C18:1cis9/C12:0+C14:0, C18:1cis9+C18:0/C12:0+C14:0, C18:1cis9/C14:0, C18:1cis9+C18:0/C14:0, C18:1cis9/sum C5:0 to C14:0, C18:1cis9+C18:0/sum C5:0 to C14:0 or C18:1cis9/C15:0. FA and their ratios alone explained 63.6% of the EB variance, indicating the possibility to use milk FA and their ratios as sole predictors for the energy status in dairy cows.

Keywords: biomarker; body condition score; energy balance prediction; fatty acids ratio; milk composition.

Figure A1

Actual weights (dots joined with grey line) and predicted daily weights (red lines) of the cows.

Figure A2

Body condition score values (dots joined with grey line) and predicted daily body condition score values (red lines) of the cows.

Figure A3

Predicted daily energy balances of the cows in effective energy (grey lines) and overall mean (bold black line).

Figure A4

Relationship between energy balance (on vertical axis) and milk fatty acid groups and ratios (on horizontal axis) modelled with a natural cubic spline. The black narrow line denotes the overall linear relationship, and the red line denotes the fitted curve. The average cross-validated root mean square error (CV RMSE) with its standard deviation is presented numerically.

Figure 1

Results of variance partitioning analysis fitted with Euler diagram. The numerical values present the percentages of energy balance variance accounted for by milk production traits (milk yield, fat and protein percentage, lactose and fat-to-protein ratio), body condition score (BCS) at calving, milk fatty acids (FA) and their ratios (FA ratios; 26 ratios in total—see Table 1) and days in milk (DIM) and their intersections.

Figure 2

Results of random forest analysis: variables’ importance according to the mean decrease in residual sum of squares from splitting on the variable (higher bar indicates higher importance).

Figure 3

Results of random forest analysis predicting energy balance in MJ of effective energy based on 26 fatty acid traits. (A) Cross-validated (CV) prediction error by cows (one line corresponds to observations of one cow) and days in milk; (B) distribution of cross-validated root mean square errors (CV RMSE) per cow, vertical dotted line with numerical value denotes the average CV RMSE.

Figure 4

Results of random forest analysis predicting energy balance in MJ of effective energy based on 26 fatty acids and ratios, days in milk (DIM), body condition score (BCS) at calving and milk traits. (A) Variables’ importance according to the mean decrease in residual sum of squares from splitting on the variable; (B) cross-validated (CV) prediction error by cows (one line corresponds to observations of one cow) and days in milk; (C) distribution of cross-validated root mean square errors (CV RMSE) per cow, vertical dotted line with numerical value denotes the average CV RMSE.

Figure 4

Results of random forest analysis predicting energy balance in MJ of effective energy based on 26 fatty acids and ratios, days in milk (DIM), body condition score (BCS) at calving and milk traits. (A) Variables’ importance according to the mean decrease in residual sum of squares from splitting on the variable; (B) cross-validated (CV) prediction error by cows (one line corresponds to observations of one cow) and days in milk; (C) distribution of cross-validated root mean square errors (CV RMSE) per cow, vertical dotted line with numerical value denotes the average CV RMSE.