Characterization and quantification of secondary metabolite profiles in leaves of red and white clover species by NIR and ATR-IR spectroscopy

Different varieties of two clover species (Trifolium pratense L. and Trifolium repens L.), cultivated in 2008 and 2009 were analysed by near-infrared (NIR) and mid-infrared (MIR) spectroscopy for establishing a fast and reliable quantification protocol for isoflavones and phenolic acids. Based on HPLC–UV/MS reference data, good results were obtained by PLS regression for the prediction of total isoflavone (R2 = 0.90) as well as for glycitin content (R2 = 0.88). Because of the lower concentration of formononetin and phenolic acids, their prediction quality was generally slightly lower (R2 = 0.73 and R2 = 0.64, respectively) compared to those of the isoflavones. The applicability of ‘leave one out’ cross validation for such a large data set is proven by comparison to an averaged randomized test-set validation leading to similar results. Additionally, the large sample set (n = 624) was screened by hierarchical cluster analysis allowing a fast evaluation of influences resulting from different cultivation parameters on the isoflavone and phenolic acid content. Climatic changes (cultivation year, date of harvest) seem to have the most impact on the metabolic profile as indicated by higher variability in the referring spectra when both cultivation years were simultaneously regarded. This work offers a new vibrational spectroscopic approach for the qualitative and quantitative determination of isoflavone and phenolic acid profiles, directly performed in the plant material without any laborious sample preparation and time-consuming chromatography. Once validated by HPLC reference, MIR and NIR spectroscopy can be used for the reliable prediction of secondary metabolites in clover as well as for fast screening and pre-evaluation of the diversity of a large sample set, aiming to reduce analytical costs, chemical waste and expenditure of time.