Multivariate analysis of milk fatty acid composition is able to discriminate dairy farm according to feeding regimen and predicted methane emission

Diet composition may change milk fatty acid (FA) profile by modulat- ing the amount of acetate and butyrate for mammary neo-synthesis and influencing the rumen biohydrogenation (BH) of dietary FA. Therefore, milk FA may be considered as a marker of the feeding reg- imen. The aim of this study was to evaluate the ability of linear dis- criminant analysis (LDA) applied on the milk FA composition to dis- criminate the feeding regimen adopted in 29 commercial dairy farms, considering the type of roughage in the diet. LDA was also applied to discriminate the farms on the basis of predicted methane emission (PME) rate (kg CH4/head and day) calculated as follow: (0.37+0.0392 GEI+0.0189 NDF–0.156 EE+0.0014 BW)/55.65, where GEI= Gross Energy Intake (Mcal/d), NDF=%NDF in the diet, EE=% ether extract in the diet, and BW= Body Weight (Kg). A stepwise LDA was applied to select the FA that best discriminated the farms on the basis of roughage used (hay, corn silage, other type of silage) or of the PME (<0.41, 0.41-0.44, 0.44-0.48, >0.48). Only variables with a R2>0.1 were retained at the end of the stepwise procedure. From the 56 initial FA detected in milk, 22 or 23 FA were retained in order to discriminate the farms according to roughage type or PME, respec- tively (P<0.001). The first canonical variable of both LDA explained more than 98% of variance. Results showed that farms adopting feed- ing regimens based on hay (no silage) were clearly discriminated from farms adopting silage as roughage basis of the feeding regi- men, moreover farms with feeding regimen containing corn silage were discriminated from farms adopting other type of silage. The absence of silage in the diet was positive related with the content of α-linolenic acid and CLA and negative related with linoleic acid con- tent in milk. Finally, LDA clearly discriminated also the four predeter- mined group of farms with different PME. High levels of PME were positively related with milk FA originated by rumen metabolism (branched FA and intermediates of rumen BH). On the contrary low levels of PME were related with milk FA originated by mammary metabolism (mainly short chain FA). LDA of milk FA was an effective method to discriminate between farms adopting feeding regimens with different roughage basis. Since this aspect may interfere also with methane emission, LDA may also effectively discriminate farms with different PME.