Rapid Classification of Turmeric Based on DNA Fingerprint by Near-Infrared Spectroscopy Combined with Moving Window Partial Least Squares-Discrimination Analysis Sumaporn KASEMSUMRAN,*† Nattaporn SUTTIWIJITPUKDEE,* and Vichein KEERATINIJAKAL**

Herbs and herbal medicines are generally safe and effective against certain diseases. Generally, they are extensively used in Asian, African and other countries. Nowadays, they are gradually increasing the demand of consumption in Europe, America and other developed countries.1 Turmeric (Curcuma longa L.) is a well-known traditional herb that is considered to be useful as medicament for various diseases. The expression of it possess antioxidant and anti-inflammatory,2–4 and can reduce cholesterol and glucose in blood,5,6 as well as betaamyloid formation.7 The most pharmacological properties of turmeric possessed from curcuminoid, which consists of three different compounds of curcumin, demethoxycurcumin and bisdemethoxycurcumin.8 Turmeric can grow very well in South and South East Asia, especially in India, China, Indonesia, and Thailand. In Thailand, turmeric is widely cultivated throughout the country. Therefore, various varieties of turmeric were found from different geographical areas, in which only a specific variety can contribute in the largest quantity of curcuminoid. If untested turmeric is employed as medicines, it can have no effects on treatment. This may lead to the mistreating or overuse of herbal medicines. With this global interest in herbs, there are concerns about the specification and efficacy of herbs. Due to the complex variety of herbs, in order to ensure the quality of clinical treatment, DNA methods have been proved to be effective for the variety identification of herbs.9–11 Nevertheless, such methods extensively require chemical reagents and analysis time. Subsequently, near-infrared (NIR) spectroscopy is a very promising technique, and has been verified to be rapid identification and quantification techniques, for various herbs and herbal medicines.12–16 In our previous studies of turmeric by NIR spectroscopy, we evolved the applicability of NIR to the quantitative analysis of curcuminoid in turmeric rhizomes and curcumin in turmeric herbal medicines, respectively.17,18 The best partial least squares (PLS) calibration model for curcuminoid was developed from the second-derivative NIR spectra in the region of 2040 – 2486 nm, founded by a wavelength-selection method, named moving window partial least squares (MWPLS). It provided the lowest standard error in the prediction (SEP) of 1.00% (w/w) and the highest ratio of the prediction to deviation (RPD) of 4.9.17 The latest study reported on the high predictive NIR model for curcumin obtained once apply of turmeric herbal medicine powder in commercial herbal medicines.18 Studies of quantitative analysis of curcuminoid in turmeric by NIR spectroscopy were also reported by Tanaka et al.19 and Kim et al.20 However, the NIR technique has not been previously reported for the classification of turmeric variety. Therefore, we expanded our study with the aim to investigate the NIR performance concerning the classification of turmeric variety based on the genotype related curcuminoid. Due to the key of 2017 © The Japan Society for Analytical Chemistry