Article pubs.acs.org/JAFC
Classification of Chinese Honeys According to Their Floral Origins Using Elemental and Stable Isotopic Compositions Zhaobin Wu,† Lanzhen Chen,*,†,§ Liming Wu,†,§ Xiaofeng Xue,†,§ Jing Zhao,*,† Yi Li,†,§ Zhihua Ye,# and Guanghui Lin⊥,Δ †
Institute of Apicultural Research, Chinese Academy of Agricultural Sciences, Beijing 100093, China Risk Assessment Laboratory for Bee Products, Quality and Safety of Ministry of Agriculture, Beijing 100093, China # Institute of Quality Standard and Testing Technology for Agro-products, Chinese Academy of Agricultural Sciences, Beijing 100081, China ⊥ Center for Earth System Science, Tsinghua University, Beijing 100084, China Δ Graduate School at Shenzhen, Tsinghua University, Shenzhen 518055, China §
ABSTRACT: The objective of this study is to test the feasibility of multi-isotopic and elemental analyses combined with chemometric techniques for differentiating the botanical origins of major honey products in China. The stable isotope and elemental compositions of 57 honey samples from four major floral origins in China (i.e., rape honey, acacia honey, vitex honey, and jujube honey) were analyzed using stable isotope ratio mass spectrometry (IRMS) and inductively coupled plasma mass spectrometry (ICP-MS), respectively. The results showed that hydrogen and oxygen isotopes could be more suitable than the carbon isotope for discriminating the floral origins of major honeys in China. There were significant differences in the contents of most elements between or among different floral origins. The combination of IRMS and ICP-MS methods provides the most effective and accurate approach (in most cases close to 100% accuracy) for classifying Chinese honeys according to their floral origins. KEYWORDS: honey, chemometric techniques, floral origins, IRMS, ICP-MS
■
techniques including physicochemical parameters,5−7 ICP-MS,8 liquid chromatography−diode array detection−tandem mass spectrometry,9,10 infrared spectroscopy,1,11,12 FT-Raman spectroscopy,13 NMR,14 gas chromatography,15 and gas chromatography−mass spectrometry7,15−17 have been developed to replace the conventional method. Stable isotope techniques have been successfully applied to detect the adulteration of honey worldwide.18−21 For instance, Ç inar et al.18 proved the C4 sugar value was a useful criteria for identifying the distinctiveness and detecting high-fructose corn syrup (HFCS) adulteration of pine honey. Stable isotope technique has also been employed to discriminate honey of differing geographical origins.20 Schellenberg et al.20 revealed that the stable isotopic compositions of honey produced in areas with geographical features and diverse climatic conditions could become reliable geographical origin information. Several previous studies have shown that elemental profiling also provides a good prospect for characterizing geographical origins of honey. As demonstrated by Pellerano et al.,22 the honey in Argentina could be divided into three producing areas according to their geographical origins by analyzing trace elements in multifloral honey from different geographical locations in combination with instrumental neutron activation analysis. Element profiling, especially using metal contents, has
INTRODUCTION Honey is known as a natural substance produced by Apis mellifera (bees) from the nectar or honeydew of plants and usually contains no additives or preservatives.1 It can be divided into various types of honey such as acacia, vitex, rape, jujube, and many more according to their sources and constitutions. Honey contains numerous nutrients and may also show biologically active properties (e.g., antimicrobial) depending on its floral origin, which is considered essential to the healthpromoting functions of honey for the human body.2 Because of the large difference in the prices and yields as well as the quantity demand among different varieties of honey, honey products have frequently been the object of fraud in Chinese and international markets. Fraudulent honey products in markets are often manufactured by combining honey resulting from various floral sources where bees forage for different nectars and honeydews, mixing premium and high-grade quality honey with cheaper and lower quality ones, or adding cheaper sweeteners. Thus, the determination and certification of honey according to their floral origins plays an important role in ensuring fairness and equality for beekeepers, producers, distributors, and consumers as well as regulatory agencies. To guarantee fair play in markets and combat product fraud, we need appropriate and reliable methods to determine the legitimacy of botanical origins of honey in the markets. Melissopalynoloy, the study of pollen components and load within honey, has conventionally been employed to determine floral origin. However, the limitations of this method include unreliability, expertise, time consumption, etc.3,4 Several new © XXXX American Chemical Society
Received: October 10, 2014 Revised: May 15, 2015 Accepted: May 20, 2015
A
DOI: 10.1021/acs.jafc.5b01576 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry
autosampler (Finnigan AS2000, Thermo Scientific). An appropriate cover on the pyrolyzer autosampler allowed continuous flushing with helium to guarantee dryness of the samples during measurement. About 1.0 g samples of honey were diluted to 10 mL with ultrapure water, and 3 μL of diluted solution was then placed into a tin capsule for 13C/12C measurements. For hydrogen and oxygen isotopic analyses, about 200 μg of fresh honey of each sample was transferred into a silver capsule in dupluicate as for the 0.2 mg each of two laboratory standards (P1, P2) with known D/H and 18O/16O values. All of these silver capsules were not completely sealed so the samples were equilibrated with laboratory vapor with stable isotopic compositions for at least 3 days before the capsules were closed with tweezers and the samples were analyzed in the TC/EA pyrolyzer for D/H and 18O/16O. The stable isotope compositions were recalculated from the IRMS results using P1 and P2 as QC materials. In this way, only the nonexchangeable hydrogen of honey was tested for D/H.30 Carbon, hydrogen, and oxygen isotope ratios were reported in δ notation in units of per mil or parts per thousand (‰) relative to the accepted international standards, δ13C‰ relative to Pee Dee Belimnite (PDB), whereas δ2H and δ18O refers to V-SMOW (Vienna Standard Mean Ocean Water). The delta values were calculated as
been used as another chemical profiling method to determine botanical origins of honey. Because metals are stable in honey over a long time, their concentrations should depend on their botanical origins, the type of soils in which nectar-producing plants grow, and the vegetation conditions in the forage area.23 For example, Chen et al.24 analyzed 12 elements in Chinese honey samples of diverse types and successfully differentiated Chinese honey through partial least-squares discriminant analysis and back-propagation artificial neural network. Chudzinska et al.25 also combined multielement concentration analysis and chemometric analysis to forecast the three different types of honey in Poland. As reported previously, the identification of geographical and botanical origins of agro-products using multistable isotope and multielement combination with chemometric analysis is an expanding field.26−29 For example, Branch et al.26 analyzed the contents of Cd, Pb, Se, Sr, and 13C, 15N for the seeds of 20 local Triticum aestivum varieties to separate different cultivars or blends using discriminant analysis. To our knowledge, no research has been performed to classify the botanical origins of Chinese honey on the basis of multistable isotope and element compositions. This study thus aimed at testing the feasibility of multi-isotopic and elemental analyses combined with chemometric techniques to differentiate the botanical origins of major honeys in China.
■
δx = (R samp/R ref − 1) × 1000 where Rsamp and Rref are the ratios of heavy to light isotope for the sample and reference material, respectively. The quality control standards for the δ13C, δ2H, and δ18O analyses were IAEA600 (δ13C = −27.771‰), USGS43 (δ2H = −50.3‰), EMA P1 (δ2H = −25.3‰; δ18O = 20.99‰), EMA P2 (δ2H = −87.80‰; 26.9‰), and USGS43 (δ18O = 14.11‰). The analytical precision was 0.2‰ for δ13C, 2‰ for δ2H, and 0.2‰ for δ18O. Multielement Analyses. The multielement analysis was performed using an Agilent 7500ce ICP-MS (Agilent Technologies Inc., Santa Clara, CA, USA). The honey samples were measured after the hot acid extraction using a heating block digestion system.31 Briefly, a ∼1.0 g sample of honey was placed into a 50 mL Teflon digestion vessel with 10 mL of HNO3 (guarantee reagent) and digested at ambient temperature overnight. The sample was placed on a heating block the next day. The combustion procedure was as follows: (1) at 70 °C for 10 min, (2) at 160 °C for 240 min, (3) at 180 °C for 120 min. The digested solution was diluted to a final volume of 50 mL with ultrapure water, decanted into luer-lock syringes, and then filtered through a Pall PVDF 0.45 μm filter into a plastic flask before further analyses. A blank digest was carried out in the same way. All materials were cleaned by soaking for 48 h in diluted HNO3 (40% v/v) and rinsed three times with ultrapure water prior to use.32 The method performances for determination of 20 elements in the honey samples were evaluated by the tests of blank values, limits of detection (LOD), and spiking recovery. The spiking recoveries for these 20 elements in honey samples range from 90.2 to 109.9%, which suggests that the performance of the method allows accurate measurements of the levels of these elements in our honey samples. Statistical Analysis. Multiple-comparison ANOVA was used to compare the differences among four honey floral origins using the SPSS software (version 17.0). Principal component analysis (PCA) and linear discriminant analysis (LDA) were applied to differentiate the floral origins of the honey samples using Matrix Laboratory (MATLAB) software (version 7.9.0). PCA minimizes the possibility of losses of the information on data matrix by reducing data set dimensionality, which can preserve most of the original variability through complex combination of their variables. The principal component (PC) could account for most of the variability in data grouping. The multivariate method was chosen to assess the stable isotope and multielement data. LDA, as a supervised technique, provided a discriminant model with respect to the descriptors’ previously defined botanical origins of honey and found a rule that allocated unknown origins of honey samples to the correct group as well as studied variable discriminate best among the types of honey. To verify the power and the stability of the model, a “leave-one-out”
MATERIALS AND METHODS
Sampling and Preparation. A total of 57 honey samples, including those originated from rape (Brassica campestris L.), acacia (Robinia pseudoacacia L.), vitex (Vitex negundo var. heterophylla Rehd.), and jujube (Ziziphus jujuba Mill. var. inermis (Bunge) Rehd.) (n = 16, 16, 12, and 13, respectively), were collected directly from beekeepers during 2012 and 2013 in different regions of China (Table 1). To avoid the effect of variations of bee species, the honey samples for this study were collected only from the bee hives occupied by Apis mellifera.
Table 1. Description of Sampling Regions for Four Major Types of Unifloral Honey in China floral source
geographical origin
characteristics
rape acacia vitex jujube
Yunnan and Sichuan provinces Beijing and Shanxi provinces Beijing and Shanxi provinces Beijing and Shanxi provinces
white color, white crystals light white color, no crystals light amber color, crystals amber color, no crystals
All honey samples were stored in the absence of light in plastic containers that were acid washed before use and maintained at room temperature (about 25 °C) until further analyses. Samples also were separated for stable isotope analysis. A test portion of about 30 g of honey was taken and liquefied at 45 °C and then homogenized in an ultrasonic bath for 0.5 h prior to both stable isotopic and multielemental analyses. Stable Isotopic Analysis. The stable isotope analyses were carried out using an isotope ratio mass spectrometer (IRMS) (Finnigan Delta V Plus, Thermo Fisher Scientific Corp, Bremen, Germany) in the IRMS Laboratory of Chinese Academy of Forestry, coupled with an elemental analyzer (EA; Flash EA1112, Thermo Scientific) to measure 13 C/12C ratios and a pyrolyzer (Finnigan TC/EA, High-Temperature Conversion Elemental Analyzer, Thermo Scientific) to measure 2 H/1H and 18O/16O ratios. To separate the gases, the EA was fitted with a Porapack QS GC column (3 m; 6 × 4 mm, o.d./i.d.) and the pyrolyzer with a molecular sieve 5 Å (0.6 m) GC column. The devices were interfaced with the IRMS through a dilutor (Conflo III, Thermo Scientific) for dosing the sample and reference gases and fitted with an B
DOI: 10.1021/acs.jafc.5b01576 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry cross-validation discriminant analysis was carried out. In this test, known samples were used as “unknowns” to validate the model built on the basis of a reduced set of cases. Effectively, one sample was removed from the data set and then classified on the basis of a model constructed from the remainder. This process was then repeated for each sample in turn and the success of the classification recalculated by comparison with the known origin.33
■
RESULTS AND DISCUSSION Variations in Stable Isotope Ratios of Honey among Floral Origins. The variations in stable isotopic compositions of the honey with different floral origins are presented in Figure 1. The order of the botanical origins according to the honey δ13C value was vitex > jujube ≈ acacia > rape (Figure 1a). Despite significant difference in the mean δ13C value among four floral origin by ANOVA test (P = 0.05), there were only slight differences in their ranges, resulting in overlapping among the floral origins. In other words, there was no clear differentiation in the δ13C values among the four Chinese honey types. The nectar sources of Chinese honey are all from C3 plants, so no big differences in the δ13C values of honey were expected due to their botanical origins. However, the differences in the δ13C values could be explained on the basis of the geographical origins and environmental conditions where the honeys were formed, but to much lesser degree on their botanical origins.20,34 For example, the rape plants tend to grow in southern and central China where the climate is more humid so their honeys have much lower mean δ13C values because of larger δ13C discrimination during photosynthesis in a humid environment,35 whereas the acacia, vitex, and jujube plants grow in drier regions, resulting in slightly higher mean honey δ13C values (Figure 1a). Thus, stable carbon ratio alone was not suitable for the determination of botanical origins for Chinese honeys. There were highly significant differences in the mean values of both hydrogen and oxygen isotope ratios according to the floral origins (all P < 0.01 from ANOVA and Duncan’s LSD tests). The order of floral origin was acacia > jujube > vitex > rape according to the mean δ2H value of honey samples (Figure 1b), and for the mean δ18O value the order was jujube > rape > acacia > vitex (Figure 1c). From a two-dimensional plot of stable hydrogen and oxygen isotope ratios (Figure 2), it was obvious that botanical origins of Chinese honeys could be separated in most cases with only these two isotopic parameters. In general, the jujube honey had high δ2H and δ18O values, so they appeared at the top right section of the plot, whereas the acacia honey had high δ2H but low δ18O values, so they clustered in the top right section, also (Figure 2). With medium δ18O but low δ2H values, the rape honeys were positioned to the middle bottom, whereas the vitex honeys were positioned in the left middle section with medium δ2H but low δ18O values, which were overlapped with several acacia or vitex honeys (Figure 2). The stable hydrogen compositions in the honey samples are determined by the isotopic compositions of the water to which the plant has assimilated and conserved during photosynthesis,20,36 whereas the compositions of stable oxygen isotopes in the honey samples were affected not only by waters available for plants but also atmospheric CO2. Because of different environmental waters and unique plant isotopic interactions with atmospheric CO2, the honeys with different floral origins could have distinct hydrogen and oxygen isotopic signals, which could be used as biological markers to discriminate the floral origins in China.
Figure 1. Box plot for carbon isotope ratio (δ13C, a), hydrogen isotope ratio (δ2H, b), and oxygen isotope ratio (δ18O, c) of Chinese honey samples according to floral origins. For each box plot, the boundary of the box closest to zero indicates the 25th percentile, the solid lines within the box mark the mean and median, and the boundary of the box farthest from zero indicates the 75th percentile. Whiskers above and below the box indicate the 90th and 10th percentiles, whereas symbols represent the 5th and 95th percentiles. Significant differences are defined at the 95% confidence level.
Comparisons in Multielement Concentrations of Honey among Floral Origins. As shown in Table 2, there was certain variability of the element compositions in the honey samples tested. Elements grouped as major elements were those having an average concentration of >1 mg/kg in food products.37 In agreement with previous studies,23,24,38 K was the most abundant chemical element in Chinese honeys (Table 2). In the honey samples tested, the highest K contents were found in the jujube honey samples, followed by the vitex, acacia, and rape honey samples. Na is the second most common chemical element in honey,23 and the order of mean Na concentration was jujube honey (46.59 mg/kg) > vitex honey C
DOI: 10.1021/acs.jafc.5b01576 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry
ratios or elemental compositions in honey samples provided only an incomplete view with respect to interpretation of honey botanical origins. Thus, it is impossible to assign the botanical origin of an unknown test sample using single-element analysis, and a multivariate evaluation is necessary to improve the overall accuracy of any classification of honey floral origin. A 3-D plot of first principal component (PC1), second principal component (PC2), and third principal component (PC3) obtained by PCA is shown in Figure 3. The plot
Figure 2. Plot of hydrogen isotope ratio (δ2H) versus oxygen isotope ratio (δ18O) for Chinese honey samples of four different floral origins.
(31.62 mg/kg) > rape honey (29.71 mg/kg) > acacia honey (22.75 mg/kg). The mineral composition of honey, especially the contents of major elements, correlated with its color. Dark and amber honeys such as the jujube honey contained higher amounts of certain major elements than pale honeys such as the rape and vitex honeys. Indeed, the color of jujube honey was darkest among the four unifloral honey types (Table 1), and the concentrations of K, Na, Ca, and Mg in the jujube honey were exactly the highest among the four botanical origins (Table 2). The contents of trace elements (usually
