Xyloglucan is a hemicellulose that occurs in the primary cell wall of all vascular plants; however, essential enzymes for xyloglucan metabolism, like XTH and β1→4-Glucan Synthase are found in Charophyceae algae.[1] In many dicotyledonous plants, it is the most abundant hemicellulose in the primary cell wall.[2] Xyloglucan binds to the surface of cellulosemicrofibrils and may link them together. It is the substrate of xyloglucan endotransglycosylase, which cuts and ligates xyloglucans, as a means of integrating new xyloglucans into the cell wall. It is also thought to be the substrate of alpha-expansin, which promotes cell wall enlargement.
Chemistry
Xyloglucan has a backbone of β1→4-linked glucose residues, most of which are substituted with 1-6 linked xylose sidechains. The xylose residues are often capped with a galactoseresidue sometimes followed by a fucose residue. The specific structure of xyloglucan differs between plant families.
Biosynthesis
Xyloglucan is synthesized in Golgi trans cisternae and in the trans Golgi network (TGN) and is transported to the cell membrane by vesicles, where it is expelled and adsorbs on nascent cellulosic microfibrils.[3]
Xyloglucan Metabolism in the human gut
The human genome doesn’t contain the genes coding for xyloglucan degradation even though xyloglucans are an important component of most human diets. Recent studies have shown that a discrete genetic locus confers xyloglucan metabolism in selected human gut Bacteroidetes. This findings reveals that the metabolism of even highly abundant components of dietary fiber maybe mediated by niche species. The metabolism of xyloglucans is the result of the concerted action of several enzymes and membrane transporters. However, given the high diversity of composition of xyloglucans from different plant sources, there is a keystone enzyme, an endo-xyloglucanase called BoGH5A, that has the ability to cleave a range of xyloglucanes to generate short xyloglucanes ready for uptake. A detailed analysis of the structure and function of the enzyme has revealed the presence of a domain called the BACON domain whose primary function in BoGH5A may be to distance the catalytic module from the cell surface and confer additional mobility to the catalytic domain to attack the polysaccharide. A broad active-site cleft engendering binding plasticity is the key feature allowing BoGH5A which allows it to accommodate a wide range of natural XyGs.
The prevalence of XyGs in the human diet suggests that the mechanism by which bacteria degrade these complex polysaccharides is highly important to human energy acquisition. Moreover, the rarity of XyG metabolism highlights the significance of Bacteroides ovatus and other proficient XyG-degrading Bacteroidetes as key members of the human gut microbial consortium.[4]
References
^LEV Del Bem and M Vincentz (2010) Evolution of xyloglucan-related genes. BMC Evolutionary Biology, 10:340, 1-17
^SC Fry (1989) The structure and functions of xyloglucan. Journal of Experimental Biology, 40, 1-11
^Moore PJ and Staehelin LA (1988). "Immunogold localisation of the cell wall matrix polysaccharides rhamnogalacturonan-I and xyloglucan during cell expansion and cytokinesis in Trifolium pratense L. - Implications for sectretory pathways". Planta174 (4): 433–445. doi:10.1007/BF00634471.
Lignocellulose refers to plant dry matter (biomass), so called lignocellulosic biomass. It is the most abundantly available raw material on the Earth for the production of biofuels, mainly bio-ethanol. It is composed of carbohydrate polymers (cellulose, hemicellulose), and an aromatic polymer (lignin). These carbohydrate polymers contain different sugar monomers (six and five carbon sugars) and they are tightly bound to lignin. Lignocellulosic biomass can be broadly classified into virgin biomass, waste biomass and energy crops. Virgin biomass includes all naturally occurring terrestrial plants such as trees, bushes and grass. Waste biomass is produced as a low value byproduct of various industrial sectors such as agriculture (corn stover, sugarcane bagasse, straw etc.) and forestry (saw mill and paper mill discards). Energy crops are crops with high yield of lignocellulosic biomass produced to serve as a raw material for production of second generation biofuel; examples include switch grass (Panicum virgatum) and Elephant grass.
Dedicated energy crops
Many crops are of interest for their ability to provide high yields of biomass and can be harvested multiple times each year. These include poplar trees and Miscanthus giganteus. The premier energy crop is sugarcane, which is a source of the readily fermentable sucroseand the lignocellulosic by-product bagasse.
Applications
Pulp and paper industry Lignocellulosic biomass is the feedstock for the pulp and paper industry. This energy-intensive industry focuses on the separation of the lignin and cellulosic fractions of the biomass.
Biofuels
Lignocellulosic biomass, in the form of wood fuel, has a long history as a source of energy. Since the middle of the 20th century, the interest of biomass as a precursor to liquid fuels has increased. To be specific, the fermentation of lignocellulosic biomass to ethanol[1] is an attractive route to fuels that supplements the fossil fuels. Biomass is a carbon-neutral source of energy: Since it comes from plants, the combustion of lignocellulosic ethanol produces no net carbon dioxide into the earth’s atmosphere. Aside from ethanol, many other lignocellulose-derived fuels are of potential interest, including butanol, dimethylfuran, and gamma-Valerolactone.[2]
One barrier to the production of ethanol from biomass is that the sugars necessary for fermentation are trapped inside the lignocellulose. Lignocellulose has evolved to resist degradation and to confer hydrolytic stability and structural robustness to the cell walls of the plants. This robustness or "recalcitrance" is attributable to the crosslinking between the polysaccharides (cellulose and hemicellulose) and the lignin via ester and ether linkages.[3]Ester linkages arise between oxidized sugars, the uronic acids, and the phenols and phenylpropanols functionalities of the lignin. To extract the fermentable sugars, one must first disconnect the celluloses from the lignin, and then use acid or enzymatic methods to hydrolyze the newly freed celluloses to break them down into simple monosaccharides. Another challenge to biomass fermentation is the high percentage of pentoses in the hemicellulose, such as xylose, or wood sugar. Unlike hexoses such as glucose, pentoses are difficult to ferment. The problems presented by the lignin and hemicellulose fractions are the foci of much contemporary research.
A large sector of research into the exploitation of lignocellulosic biomass as a feedstock for bio-ethanol focuses particularly on the fungus Trichoderma reesei, known for its cellulolytic abilities. Multiple avenues are being explored including the design of an optimised cocktail of cellulases and hemicellulases isolated from T. reesei, as well as genetic-engineering-based strain improvement to allow the fungus to simply be placed in the presence of lignocellulosic biomass and break down the matter into D-glucosemonomers.[4] Strain improvement methods have led to strains capable of producing significantly more cellulases than the original QM6a isolate; certain industrial strains are known to produce up to 100g of cellulase per litre of fungus[5] thus allowing for maximal extraction of sugars from lignocellulosic biomass. These sugars can then be fermented, leading to bio-ethanol.
Published Date May 2016, Vol.84:38–43,doi:10.1016/j.vibspec.2016.02.011 Title Analytical method development using FTIR-ATR and FT-Raman spectroscopy to assay fructose, sucrose, glucose and dihydroxyacetone, inLeptospermum scopariumnectar
Author
Elizabeth M. Nickless a
Stephen E. Holroyd b
Georgie Hamilton a
Keith C. Gordon c
Jason J. Wargent a,,
aInstitute of Agriculture & Environment, Massey University, Palmerston North, New Zealand
bFonterra Co-operative Group, Fonterra Research & Development Centre, Palmerston North, New Zealand
cDepartment of Chemistry and Dodd-Walls Centre, University of Otago, Dunedin, New Zealand
Received 7 September 2015. Revised 22 February 2016. Accepted 27 February 2016. Available online 3 March 2016. Abstract The carbohydrate dihydroxyacteone (DHA) occurs in significant levels inLeptospermum scoparium(mānuka) nectar and is the precursor of methylglyoxyl (MGO), the unique non-peroxide antibacterial activity (NPA) component in mānuka honey. The nectar of ten different cultivars ofL. scopariumwas assayed quantitatively for fructose, glucose, sucrose, and DHA with high pressure liquid chromatography (HPLC) for comparison with FT-Raman and FTIR-ATR spectroscopic methods. FT-Raman spectroscopy and ATR-FTIR spectroscopy, alongside chemometric methods using principal component analysis (PCA) and partial least squares (PLS) prediction were shown to be useful techniques to quantify and compare the nectar composition in a range of cultivars ofL. scoparium. Keywords
Dihydroxyacetone
Leptospermum scoparium
FT-Raman
FTIR-ATR
Nectar composition
1 Introduction
The nectar composition of Leptosperum scoparium (mānuka), as the plant nectar source for mānuka honey, is of significant interest to the honey industry in New Zealand. The nectar of L. scoparium contains the carbohydrate dihydroxyacetone (DHA) which is the precursor chemical for methylglyoxyl (MGO), the unique non-peroxide antibacterial activity (NPA) component of mānuka honey [1], [2] and [3]. Leptospermum honeys are valued for their therapeutic application in wound healing of skin infections [4].
The traditional method of applying the anthrone assay and colorimetric analysis [5] to measure total sugars in L. scoparium nectar is a time consuming and laboratory-intensive method. This study presents the potential of attenuated total reflectance (ATR) Fourier transform Infrared (FTIR) and Fourier transform (FT)-Raman spectroscopy combined with the chemometric tool Partial Least Squares Regression analysis (PLSR) to develop a method suitable for the rapid non-destructive discrimination of nectar composition, in particular DHA and saccharide sugars in L. scoparium.
Most chromatographic techniques are based on solvent extraction followed by high performance liquid chromatography (HPLC) separation and/or colorimetric and enzymatic analysis [5] and [6]. Traditional analytical HPLC methods for quantification of analytes of interest require expensive and extensive sample preparation, with additional derivatisation techniques for small molecules such as DHA [7]. FTIR techniques have been shown to be useful in analysing and quantifying sugars in a range of samples such as cereal, baby foods, honey and fruit juices during product processing [6], [8], [9], [10] and [11]. Sultanbawa et al. recently demonstrated that mid infrared techniques provided a good model for predicting methylglyoxyl levels in Leptospermum polygalifolium honeys [4]. Infrared spectroscopy is useful in identifying compounds based on the vibration frequencies of their molecular structure and is a sensitive technique for analysing the chemical composition of various sample types [6]. The application of spectroscopic techniques for the quantification of carbohydrates in-situ in plants has received some previous attention from other authors. Past work includes analysis of monosaccharides and polysaccharides applying either FTIR or Raman, and various chemometric analysis techniques to differentiate individual carbohydrates or fingerprint sample types [9]. Raman spectroscopy and FTIR spectroscopy have been used to quantify fructose, glucose, sucrose and saccharose in fruit juices and fermentation stages in vinegar and pineapple juice along with quantifying various sugars and acids for grading fruit quality [6], [9], [10] and [11]. Other findings have been reported using FT-Raman to quantify sugars in carrot tissue along with carotenoids and polyacetylene [12], [13]and [14]. FT-Raman has also been used to analyse the chemical composition of floral honey [15]. Both FTIR and FT-Raman analytical spectroscopy methods are widely applied to identify substances from characteristic spectral patterns (“fingerprinting”) and to ascertain quantitatively or qualitatively the amount of a substance in a sample [16] and [17]. Infrared spectroscopy has been used successfully in a preliminary study on nectar using principal component analysis to distinguish various different plant species nectars [18]. To date there has been no research published using these methods to quantify individual plant nectar components.
2 Experimental methods
The floral nectars of ten different proprietary cultivars of L. scoparium were investigated using both attenuated total reflectance (ATR), Fourier transform Infrared (FTIR), and Fourier transform (FT)-Raman spectroscopic techniques, along with high pressure liquid chromatography (HPLC) as a reference analysis. Each cultivar set contained ten replicate plants grown from cuttings of a single parent plant and so were genetically identical i.e. clones of the parent material. Plants were supplied by Comvita New Zealand, and were grown in standard tree and shrub potting mix in 30 cm pots in a glasshouse at the Plant Growth Unit at Massey University, Palmerston North, New Zealand. Nectar was collected in a consistent fashion from 20 flowers from each of the 10 plants of each cultivar and was pooled. Nectar was collected at randomly selected times between 10am and 2pm on any day from flowers at development stage IV [19]. DHA levels in nectar samples were measured using aqueous extraction, derivatisation and analysis by HPLC adapting the method used by Windsor et al. to analyse DHA in honey samples [7]. In addition, DHA levels in nectar samples were validated using a commercial analytical service (R.J. Hills Laboratories Hamilton, New Zealand: http://www.hill-laboratories.com/page/pageid/2145845743/Honey_Testing). HPLC analysis of the sugars fructose, glucose, and sucrose were performed at Institute of Agriculture & Environment, Massey University; Palmerston North, New Zealand.
Three replicates from each nectar sample were used for the development of the spectroscopic methods. ATR-FTIR was used to analyse the same components and quantify using Partial least Squares regression (PLSR), and results and error values were compared to test the accuracy of using FTIR. FT-Raman spectra from a separate nectar set was collected for comparative DHA quantification capability.
2.1 HPLC conditions for DHA analysis
Analyses were performed on a PerkinElmer Series 200 Pump and Auto sampler with a Flexar photo diode array detector (λ = 263 nm). HPLC separations were performed on a Synergi Fusion column (75 × 4.6 mm, 4 μm particle size). The column was heated and held at 30 °C to maintain stable run conditions. Mobile phase A was water: ACN, 70/30, v/v and mobile phase B was 100% ACN. The following 23 min gradient elution was employed: A:B = 90:10 (isocratic 2.5 min), graded to 50:50 (8.0 min), graded to 0:100 (1.5 min), 0:100 (isocratic 7.0 min), graded to 90:10 (1.0 min), 90:10 (isocratic 3.5 min), detection at 263 nm.
2.2 Preparation of reaction solutions
Hydroxyacetone (HA) (3.01 mg/ml) formed the HA internal standard solution. The O-(2, 3, 4, 5, 6-pentafluorobenzyl) hydroxylamine (PFBHA) derivatising reagent was 19.8 mg/ml in citrate buffer (0.1 M) adjusted to pH 4 with sodium hydroxide (NaOH) (4 M). DHA (3.88 mg/ml) formed the DHA standard solution.
2.3 Sample preparation
For the preparation of standards, DHA standard solution (100, 80, 60, 40, 20, 10 and 0 μl) was added to tubes 1–7 respectively, and made up to 100 μl with nanopure water. For sample analysis, 20 μl of nectar or standard was pipetted into a mix tube and 25 μl of the HA was added. Derivatisation steps were performed at 25 °C in a controlled temperature room. Each of the HPLC samples and standards was thoroughly mixed and placed in a rack on a rotating Table for 1 h to allow complete dissolution. PFBHA derivatising solution (100 μl) was added to each test tube, which was mixed and placed in a rack on a rotating Table for 1 h to allow for complete derivatisation. Acetonitrile (ACN) (1.5 ml) was added to each test tube and mixed. Nanopure water (0.5 ml) was then added to each test tube and mixed. Samples were then syringe filtered with a 0.22 μm filter into HPLC vials. Vials were placed into the auto sampler and run overnight, and repeat analysis of standards were analysed through each run to check stability of the analysis. DHA calibration curves were generated from tubes 1–7 by linear regression using the HPLC peak area ratios of DHA: HA plotted against the mass of the DHA mass content of the nectar samples were determined against these calibration curves.
2.4 HPLC conditions for sucrose, glucose and fructose analysis
Individual sugar analysis was performed on an Alliance Waters 2690 Separations Module HPLC and auto sampler using a Waters 2410 Refractive Index (RI) detector.
Sucrose, glucose and fructose were separated using an Aminex HPX-87C Carbohydrate Column (300 × 7.0 mm). The column was heated and held at 65 °C to maintain stable conditions. A standard run method was used with one phase solution of milliQ water at a flow rate of 0.6 ml/minute. The sample injection size was 20ul and each sample run was 18 min.
2.5 Preparation of standards
1% stock solutions of each of the sugars sucrose, glucose, and fructose were made in nano-pure water. Five standards were used to create a calibration curve for each sugar; the standards contained equal proportions of sucrose, glucose and fructose, and had concentrations of 0%, 0.01%, 0.05%, 0.17%, and 0.33% sugar.
2.6 Sample preparation
30 μl of each nectar sample was added to 1470 μl of nano-pure H2O and mixed. The diluted nectar samples as well as the standards were filtered using 0.22um syringe filters into HPLC vials. Vials were loaded into the auto-sampler along with the standards and analysed overnight. Standards were repeat assayed throughout the run to reference for any variability during the analysis.
HPLC analysis of the sugars fructose, glucose and sucrose was performed at Institute of Agriculture & Environment, Massey University; Palmerston North, New Zealand.
2.7 Spectroscopic methods
For FTIR analysis of nectar, three replicates from each standard were analysed using a Bruker Tensor 37 MICRO-ID mid-IR source FTIR (Ettlingen, Germany). Samples were scanned at a resolution of 4 cm−1 and a scan average of 64 scans; wavelengths scanned 4000–0 cm−1. Five micro-litres of each sample were placed on the zinc selenide ATR crystal for scanning. FTIR Spectra were analysed using PCA and PLS methods using statistics software Unscrambler V10.2 (CAMO, Norway).
For FT-Raman analysis of nectar, samples were pipetted into aluminium divots and were analysed by Raman spectroscopy on a Bruker Equinox 55 interferometer equipped with the Bruker FRA 106/S FT-Raman accessory (Ettlingen, Germany) and a D418 T liquid-nitrogen-cooled germanium detector was used, controlled by the Bruker OPUS v6.0 software and with a Nd:YAG 1064 nm excitation source. A 1 mm diameter laser spot size and 400 mW power setting were used, 120 scans per spectrum with a spectral resolution of 4 cm−1 in the wave-number range 3500−0 cm−1 were produced. Samples were analysed in triplicate. Raman spectra were analysed using PCA and PLS methods using statistics software Unscrambler V10.2 (CAMO,Norway).
Reference spectra of 8% DHA-phosphate made up in milliQ H2O and a mock nectar reference solution of fructose:glucose 2:1 were analysed with both ATR-FTIR and FT-Raman to compare spectral data between the two analytical methods.
3 Results and discussion
The DHA, fructose, glucose, sucrose, and total sugar values obtained from the nectar of the cultivars using HPLC are shown in Table 1. Results show that the composition of L. scoparium nectar consists mainly of the monosaccharides fructose and glucose, in agreement with previous research [20], and also low concentrations of the disaccharide sucrose (see Table 1). L. scoparium also contains the triose sugar dihydroxyacetone (DHA) in significant concentrations [21], [22], [23], [24], [25] and [26]. Statistical analyses of the comparison of means of each cultivar show that there are five significantly different groups in regard to DHA concentration in the ten cultivars analysed, along with significant differences in monosaccharide and disaccharide composition.
Table 1. HPLC data showing DHA and sugar component values for the cultivars normalised to 80°BRIX. Error statistics show standard error of the means. Letters adjacent to column values in superscript indicate the results from the analysis of the comparison of means of each cultivar applying ANOVA with a Tukey pair-wise comparison of the cultivar lines using Minitab statistical software, same letter indicates no significant difference between the cultivars at the 95% confidence level and P-values ≤ 0.05.
Cultivar
Fruct% 80°BRIX
S.E.
Gluc% 80°BRIX
S.E.
Sucrose% 80°BRIX
S.E.
DHA 80°BRIX mg/kg
S.E.
BS
43.22 f
0.22
35.94 a
0.19
0.85 b
0.06
5177.02 b,c,d,e
677.10
G
44.90 e,f
0.73
33.24 b
0.59
1.87 a
0.20
4116.01 a,b
521.43
O
46.73 d,e
0.51
32.82 b,c
0.49
0.45 c,d
0.04
5981.54 b,c
442.12
PU
46.87 c,d,e
0.27
32.47 b,c,d
0.25
0.66 b,c
0.02
2714.35 b,c,d
380.44
B
46.97 b,c,d,e
0.45
32.68 b,c,d
0.41
0.35 c,d
0.07
6153.12 e
469.22
R
48.93 a,b,c,d
0.40
30.78 c,d,e
0.37
0.29 d
0.05
7459.07 a,b
373.56
P
49.25 a,b,c
0.46
30.55 d,e
0.40
0.20 d
0.07
3058.88 c,d,e
352.04
Y
49.41 a,b
0.73
29.90 e
0.73
0.69 b,c
0.02
4843.71 d,e
111.35
MG
49.70 a
0.86
29.79 e
0.84
0.51 b,c,d
0.02
3266.70 a
554.31
LG
50.25 a
0.20
29.51 e
0.20
0.24 d
0.01
4252.20 b,c,d,e
366.02
The saccharide data was normalised against the total sugar content to allow comparisons of the composition between the cultivars. Principal component analysis of the HPLC data of the saccharide nectar sugars glucose, fructose and sucrose show significant groupings according to cultivar (CV) type (Fig. 1A). In this plot, cultivars are grouped and spread mainly along PC1, which is influenced by glucose and fructose concentrations. Cultivars with higher fructose concentration are shown towards the positive end of PC1, and higher glucose concentration towards the negative end, as shown by the loadings for PC1 in Fig. 1B. Cultivar G has a significantly higher sucrose concentration compared to the other nine cultivars, as highlighted in the PC plot with CV G in a separate grouping along PC2 compared to the position of the other CV groups along PC2. PC2 is mainly influenced by sucrose concentrations as indicated by the loadings plot in Fig. 1B. Regression analysis of the two mono-saccharides (fructose against glucose) showed a negative correlation, indicating that as fructose concentration increases, the glucose concentrations decrease in the nectar of L. scoparium. These data suggest an equilibrium not due to random effects within the sugar metabolism of floral nectar in L. scoparium.
Fig 1. Principal component analysis score plots using the main sugar composition (fructose, glucose and sucrose) of the nectar of the ten cultivars: B, BS, G, LG, MG, O, P, PU, R and Y. A = PCA score plot B = Loadings plot.
FTIR and FT-Raman spectral signatures for the saccharides fructose, glucose, and the sucrose are well documented [6], [9], [13] and [15]. However there is a paucity of literature describing the spectral signature for the monosaccharide DHA. The Raman spectra of DHA has a distinct peak at 1740 cm−1 from the carbonyl bond v(CO) [13] (Fig. 3), which is not present in other saccharides in the floral nectar of L. scoparium (Fig. 2B). However, at low concentrations of DHA, as are present in nectar assayed, this peak is not as distinct. The 1740 cm−1 peak is visible as a smaller peak in the Raman spectra in nectars of L. scoparium (Fig. 2B), where the DHA concentrations are higher; for example, Cultivars O and B (Table 1). Other peaks typical of saccharide sugar bonds are observed e.g. CO stretch vibrations at 1077 cm−1 and 1062 cm−1, C CH, O CH and C OH deformations at 1268 cm−1, 1124 cm−1 and 915 cm−1, CH2 assymetric and symmetric modes at 1457 cm−1, 2946 cm−1 and 1365 cm−1, CH deformation at 777 cm−1 and 915 cm−1. With various skeletal modes C CO and C CC at 524 cm−1 and 426 cm−1 and ring deformations showing as a weak band at 624 cm−1[15].
Fig. 2. Comparison of FTIR (A) and FT-RAMAN (B) spectra of 8% DHA, L. scoparium nectar and a composition of mock nectar made from the saccharide solutions fructose and glucose (ratio 2:1).
Fig. 3. DHA regression model plots from FTIR-ATR (A) and FT-Raman (B) spectral data from nectar samples. Also regression plots of predicted data using the predicted data from the models in A (FTIR-ATR) and B (FT-Raman) against HPLC DHA data for FTIR-ATR (C) and FT-Raman (D).
The FTIR-ATR spectral signature for DHA is much less distinct (Fig. 2A). In the FTIR spectrum broad peaks are visible underneath the main distinct peaks representing the various (CO), (CC) and (OCH) bonds present in sugars.
3.1 Regression analysis
For FTIR spectra the wavenumber region 1650–700 cm−1 was used for regression analysis. PLSR analyses were cross validated with the leave-one-out method [27]. Spectra were either used raw or processed to the 1st or 2nd derivative depending on the sugar being analysed for quantification. For FT-Raman, the region 1800–300 cm−1was used. Standard spectral pre-processing was applied using Savitsky-Golay smoothing, SNV and linear baseline applications prior to partial least squares regression (PLSR). PLS regression models for both FTIR-ATR and FT-Raman were generated using spectral data against HPLC quantification data for each sugar (non-normalised) along with total sugars. The predictive models generated from the FTIR spectral data gave excellent correlations at ranges 0.88–0.99 R-squared values (Table 2). Sucrose and DHA are in much lower concentrations in the floral nectar, and the spectral data required further pre-processing to improve the model. Sucrose required 1st order derivation and DHA data also required 1st order derivation, both with a polynomial order of two of the spectra to optimise the models for these two sugars. The prediction performance was evaluated by mean squares error (RMSE) of calibration (RMSEC) and prediction (RMSEP) [28].
Table 2. PLSR models and Regression analysis results for nectar sugars in L. scoparium using ATR-FTIR and FT-Raman.
Sugar
vs
Method
Pre-processing methods used
Standard method applied to all spectra = SG smoothing, SNV and linear baseline
RMSEC|RMSEP
R2 Calibrated/Predicted
Slope
Bias
R2 Model prediction vs HPLC
RPD
Fructose FTIR
Standard method only
0.96|1.09
0.98/0.97
0.98/0.97
−4.2/−0.02
0.97
6.17
Glucose FTIR
Standard method only
1.02/1.35
0.97/0.95
0.97/0.95
−7.3/0.05
0.95
6.7
Sucrose FTIR
Standard method plus 1st order derivation
0.06|0.11
0.94/0.83
0.94/0.80
−4.8/−8.5
0.77
1.2
Total Sugars FTIR
Standard method only
1.14|1.17
0.98/0.98
0.98/0.98
1.0e−06/0.016
0.95
9.4
DHA FTIR
Standard method plus 1st order derivation
264.52 mg/kg|357.28
0.89/0.79
0.89/0.84
−1.1/12.3
0.86
2.9
DHA FT-Raman
Standard method plus 1st order derivation
174.4 mg/kg|330.3
0.92/0.76
0.92/0.72
−6.1/−22.1
0.85
3.1
FT-Raman techniques used to analyse the nectar for quantification of DHA demonstrated that the improved contrast, with additional peaks, in Raman spectra could generate a more accurate model for DHA quantification (Table 2). Using FT-Raman, better results were provided for analysing DHA with a lower calibration error (RMSEC) from the model compared to FTIR, with an R2 of 0.98 and RMSEP of 397.62 mg/kg sugar compared with R2 of 0.88 and RMSEP of 357.28 mg/kg when using FTIR spectra (Fig. 4). Typically larger differences indicate a less reliable model. These error levels compare favourably with those obtained with HPLC techniques which gave a means standard error range of 111.35–677.10 mg/kg depending on the cultivar nectar analysed. The error between the calibration error and prediction error for the Raman model is much larger than the differences in the FTIR model, this indicates that although the regression model for FT-Raman is more accurate; the model has a larger chance of predicting errors, contrasted with the FTIR model which has a less accurate model but less chance of prediction-based errors. However, on analysis of the predicted data set from the models, outlier samples such as MG-4, O-6, MG-5 and P-5 can be detected from the method via larger standard deviation values and can be removed from the dataset (Fig. 4). Outliers could likely be due to interference from possible contaminants in the nectar such as nectar pigmentation, or possibly pollen contamination.
Fig. 4. Prediction data plot from the FT-Raman regression model. Note samples with a larger standard deviation represented by the box around the mean are considered outliers in the prediction i.e. MG-4, O-6, MG-5 and P-5. Removal of these outliers gives a regression model of FT-Raman predicted values against HPLC data with an R2 = 0.85 (Table 2).
Further analysis using predicted values from the PLSR models against the original HPLC data values for some samples was performed. Table 2 shows the regression results of the spectral analysis using PLS and the regression results showing predicted data from the PLS models against the HPLC data. R-squared values are very good for all nectar components measured excepting the sucrose model at R2 = 0.77. The R2 for the FT-Raman predictions against HPLC data are very good (0.85 R2) after removing outliers. FTIR prediction data against HPLC data was also good (0.86 R2). Overall an accurate model is possible for quantifying DHA concentrations in nectar using both FTIR-ATR and FT-Raman spectroscopy.
In evaluating the model data against raw data for DHA, the range of standard error of the means (±111.35 to ±677.10 mg/kg DHA) is quite large for nectar samples, yet the prediction errors using spectral data with PLSR methods is well within the range of intra-plant variation in terms of nectar composition. Both ATR-FTIR and FT-Raman spectroscopy can be used as a quantitative model for screening L. scoparium nectar for DHA. Fructose, glucose, sucrose, and total sugars can be also accurately quantified using ATR-FTIR techniques, alongside multivariate analysis models using PLS. The results suggest that ATR-FTIR is a sensitive technique to analyse the sugars in the nectar of L. scoparium, including DHA.
4 Conclusion
It is of benefit to the mānuka honey industry to access facile methods such as analytical spectroscopy techniques, in combination with chemometrics, to screen for DHA in the nectar of L. scoparium, along with related applications to quantify nectar components of interest rapidly and non-destructively. ATR FTIR and FT-Raman techniques have both proven to have strong potential to be of use in this application.
Acknowledgements
This work was supported by a Primary Growth Partnership (Ministry of Primary Industries, New Zealand) awarded to Mānuka Research Partnership (NZ) Limited.
O) 
O stretch vibrations at 1077 cm−1 and 1062 cm−1, C
