Discrimination of honey of different floral origins by a combination of various chemical parameters

Highlights

• Discrimination of honeys with different floral origins was achieved.

• Classification models (OPLS-DA) were successfully validated using external data sets.

• The most influential variables responsible for discrimination were identified.

Abstract

Honey is a high value food commodity with recognized nutraceutical properties. A primary driver of the value of honey is its floral origin. The feasibility of applying multivariate data analysis to various chemical parameters for the discrimination of honeys was explored. This approach was applied to four authentic honeys with different floral origins (rata, kamahi, clover and manuka) obtained from producers in New Zealand. Results from elemental profiling, stable isotope analysis, metabolomics (UPLC-QToF MS), and NIR, FT-IR, and Raman spectroscopic fingerprinting were analyzed. Orthogonal partial least square discriminant analysis (OPLS-DA) was used to determine which technique or combination of techniques provided the best classification and prediction abilities. Good prediction values were achieved using metabolite data (for all four honeys, Q² = 0.52; for manuka and clover, Q² = 0.76) and the trace element/isotopic data (for manuka and clover, Q² = 0.65), while the other chemical parameters showed promise when combined (for manuka and clover, Q² = 0.43).

Introduction

In recent years, there has been growing interest in verifying the floral origin of honey, especially in the characterization of unifloral honeys, which are often more valuable than polyfloral honeys. Certain types of unifloral honey have claimed or apparent benefits for human health and are used in the treatment of wounds and diseases because of their healing and antibacterial properties (Allen et al., 1991, Robson et al., 2009, Viuda-Martos et al., 2008; ftp://ftp.fao.org/docrep/fao/012/i0842e/i0842e00.pdf). Several studies have proven the antimicrobial, anti-inflammatory, antimutagenic, antitumor, antioxidative activities, as well as many other benefits for human health, of compounds such as phenolic acids and flavonoids (Cushnie and Lamb, 2005, Havsteen, 2002) that are components of these honeys. New Zealand manuka (Leptospermum scoparium) honey, for example, has been proven to have non-peroxide antibacterial activity (Allen et al., 1991). According to the Codex Alimentarius Standard for Honey (2001) and the European Commission Directive (2001), the use of botanical designation is allowed if a honey originates predominantly from the indicated floral source. Adulteration in terms of the dilution of honeys of high value floral origin with those of lower value has increased in recent years. Therefore, discrimination of honey by floral origin is of great importance.

Identification of the floral origin of honey is typically achieved by melissopalynological analysis based on pollen characterization, and presently is complemented by sensory and physico-chemical analysis. However, pollen identification requires a high degree of skill and in some cases gives erroneous results (Cavazza et al., 2012, Molan, 1998), while the determination of physico-chemical parameters is broad and cannot be uniformly applied to all honey varieties. For example, a citrus honey was shown to have 18% pollen content from kiwi fruit, which does not produce nectar (Moar, 1985). This clearly shows that honeys can incorporate pollen that is unrelated to the nectar source. Therefore, there is an ongoing need to develop reliable, practical, and faster methods to discriminate between honeys of different floral origins.

Recently, there has been an increase in the number of analytical techniques applied to differentiate honeys, for example by the analysis of flavonoids (Chan et al., 2013, Trautvetter et al., 2009), amino acids (Kečkeš et al., 2013, Rebane and Herodes, 2008), proteins (Wang et al., 2009), phenolic compounds (Cavazza et al., 2012, Stephens et al., 2010), honey volatiles (Senyuva et al., 2009, Stanimirova et al., 2010), carbohydrates (Senyuva et al., 2009), and trace elements (Marghitas et al., 2010). Vibrational spectroscopy techniques (NIR, FTIR, and NMR) combined with chemometrics have also been used for the determination of the floral origin of honey and development of classification models (Etzold and Lichtenberg-Kraag, 2008, Liang et al., 2013, Schievano et al., 2012). Typically, principal component analysis (PCA) was applied as a clustering method for preliminary evaluation of the data structure, followed by various classification methods such as linear discriminant analysis (LDA), discriminant partial least square regression (DPLS), soft independent modeling of class analogy (SIMCA), or back propagation neural networks (BPNN) (Etzold and Lichtenberg-Kraag, 2008, Liang et al., 2013, Schievano et al., 2012).

The high commercial value of some New Zealand honeys (L. scoparium (manuka), Kunzea ericoides (kanuka), and Trifolium spp. (clover)) has motivated intensive investigation by different research groups, their characterization being mostly based on targeted analysis of extractable organic components (Senanayake, 2006, Stephens et al., 2010, Yao et al., 2003).

To the best of our knowledge, most of the studies to date have employed a single technique (sensory, microscopic, chromatographic, or spectroscopic) applied to one group of samples. There have been no studies published on the use of combined techniques to gain additional information and to find out the most appropriate methodology for honey floral origin discrimination. The aim of this study was to explore the feasibility of a multivariate approach, applying multivariate analysis to the results of sample analysis by a number of physico-chemical techniques, for the discrimination of various honeys. This approach was applied to four authentic honeys with different floral origins (rata, kamahi, clover and manuka) obtained directly from honey producers in New Zealand. The data processing was performed on the results of chemical analyses using elemental profiling, stable isotope measurements, metabolomics (ultra-performance liquid chromatography quadrupole time of flight mass spectrometry (UPLC-QToF MS)), and vibrational spectroscopy (near-infrared (NIR), fourier transform infrared (FT-IR), and Raman spectroscopy) fingerprinting. Orthogonal partial least squares projections to latent structures discriminant analysis (OPLS-DA) was used to determine which technique or combination of techniques provided the best classification and prediction ability.