oiof

rtat P. Bo

Honey

phyd foct oveloriks t

tin, quercetin, galangin and particularly p-coumaric acid (in honeydew honey). The impact of industrialthermal treatments is lower than the expected variability as a consequence of the year of collection,

the mdetermstry, rnd qu

Therefore, it could be interesting to authenticate honey botanic

proteins, organic acids (Mato, Huidobro, Simal-Lozano, & Sancho,2006), sugars (Cavia, Fernndez-Muio, Huidobro, Alvarez, &Sancho, 2009), avonoids and other phenolic compounds havebeen widely used as a tool for this purpose (Baltruaityte, Riman-tas-Venskutonis, & Ceksteryte, 2007; Escriche, Kadar, Juan-Borrs,& Domenech, 2011; Frankel, Robinson, & Berenbaum, 1998; Iurlina,

However, it is important to emphasise that honey is not normallyit usually under-cause conugh recen

origin (botanical and geographical), temperature, moisture content,and sugar content (Cavia et al., 2009). Industrial manufacturing ofhoney includes two stages that involve thermal treatments: lique-faction (approx. 55 C) to ensure that it is sufciently liquid to han-dle; and pasteurisation (approx. 80 C) to destroy yeast that cancause unwanted fermentation during the products shelf-life anddissolve the crystallisation nuclei that cause honey to solidify,ensuring that the honey stays in its liquid form for as long as possi-ble (Escriche, Visquert, Carot, Domnech, & Fito, 2008).

Corresponding author. Tel.: +34 963879146; fax: +34 963877369.

Food Chemistry 142 (2014) 135143

Contents lists available at

he

lseE-mail address: iescrich@tal.upv.es (I. Escriche).ey.Amongst the different analytical possibilities: volatile com-pounds (Escriche, Visquert, Juan-Borrs, & Fito, 2009), metals,

vested raw honey is in a liquid state, it crystallises with greater orlesser speed. Crystallisation depends on numerous factors such assource considering specic chemical compounds that are onlypresent in specic nectars (or secretions of plants in the case ofhoneydew honey), and consequently in the corresponding hon-

commercialised in its raw state; on the contrarygoes an industrialisation process. This is mainly bedemand a uid, non-crystallised product and altho0308-8146/$ - see front matter 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.foodchem.2013.07.033sumerstly har-the identication of the botanical source of honey is carried out bythe calculation of the percentage of pollen. However, this analysisis not the most suitable for some kinds of honey such as citrus,since the amount of pollen present is generally small and very var-iable as the maximum secretion of nectar does not coincide withanther maturation (Ferreres, Giner, & Tomas-Barbern, 1994).

as markers in the determination of a specic honey: kaempferolin rosemary honey (Toms-Barbern, Martos, Ferrerer, Radovic, &Anklam, 2001), abscisic acid in heather honey (Ferreres, Andrade,& Toms-Barbern, 1996), hesperetin in citrus honey (Ferreres,Garcia-Viguera, Toms-Lorente, & Toms-Barbern, 1993), luteolinin lavender, quercetin in sunower honey (Yao et al., 2003).FlavonoidsPhenolic acidsAntioxidant capacityAuthentication

1. Introduction

Botanical authentication is one ofhoney quality control, as it directlyRegulatory authorities, the food induare interested in knowing the origin athough neither factor has enough inuence to alter these constituent compounds to the point of affectingthe discrimination of honey by botanical origin.

2013 Elsevier Ltd. All rights reserved.

ost important issues inines the market price.etailers and consumersality of honeys. Usually,

Saiz, Fritz, & Manrique, 2009; Lachman, Orsk, Hejtmnkov, &Kovrov, 2010b; Schramm et al., 2003; Yao et al., 2004). This isdue to the fact that these compounds are stable non-volatile sec-ondarymetabolites, which appear to be relatively unaffected byenvironmental factors (Ferreres, Giner, & Tomas-Barbern, 1994).For this reason, some of these compounds have been suggestedKeywords:honey); kaempferol, chrysin, pinocembrin, caffeic acid and naringenin (in rosemary honey) and myrice-Analytical Methods

Suitability of antioxidant capacity, avonfor oral authentication of honey. Impact

Isabel Escriche a,, Melinda Kadar b, Marisol Juan-Bora Institute of Food Engineering for Development, Food Technology Department, Universib Institute of Food Engineering for Development, Universidad Politcnica de Valencia, P.O

a r t i c l e i n f o

Article history:Received 1 April 2012Received in revised form 20 September2012Accepted 7 July 2013Available online 15 July 2013

a b s t r a c t

Total antioxidant activity,compounds were evaluateof Spanish honey, the imparus honey had the lowest lethe highest ones. Botanicalpermit discrimination than

Food C

journal homepage: www.eds and phenolic acidsindustrial thermal treatment

s b, Eva Domenech a

olitcnica de Valncia, P.O. Box 46022, Valencia, Spainx 46022, Valencia, Spain

sicochemical parameters, and the prole of avonoids and phenolic acidr: their ability to distinguish between the botanical origins of four typesf industrial thermal treatment, and the effect of the year of collection. Cit-s of all the analysed variables, then rosemary and polyoral, and honeydewgin affects the prole of avonoids and phenolic compounds sufciently too the predominance of particular compounds such as: hesperetin (in citrus

SciVerse ScienceDirect

mistry

vier .com/locate / foodchem

impact on the different components of honey such as: enzymes,

2.1. Honey samples and their classication

emisFour types of Spanish honey (harvested in 2008 and 2009):three of oral origin (citrus, rosemary and polyoral) and one fromhoneydew (forest origin) were used in this study. The botanical ori-gin of the samples was ascertained by melissopalynological analy-sis (Table 1). Conductivity is also given in this table for honeydewin order to supplement the pollinic information for this type ofhoney as required (Council Directive, 2002). These honeys are themost consumed in Spain. For each type of honey, and for each year,4 different raw batches (15 kg each) were obtained directly fromlocal beekeepers (to ensure freshness). Each batch was divided into3 parts; one was analysed in its raw state (unheated) and the other2 were analysed following heat treatment (liquefaction and lique-faction plus pasteurisation). Industrial liquefaction and pasteurisa-tion processes were taken into account when choosing thetreatment temperatures. Accordingly, the liquefaction sampleswere placed in a temperature-controlled oven (Selecta model20000207 80L, Barcelona, Spain) at 45 1 C for 48 h, and the pas-teurisation samples (previously liqueed) were heated at80 0.05 C for 4 min in a temperature-controlled oil bath (Digi-term 200, Selecta, Barcelona, Spain). The honey was pumped, usinga variable speed peristaltic pump (Heidolph model Pumpdrive5001, Schwabach, Germany), through silicone tubing (6 mm boreand 1 mm thickness) in the bath. Given that not all the honeyshad the same viscosity, the pump speed was adjusted in each caseto achieve the desired pasteurisation time. After thermal treat-ments, all samples were quickly cooled to 30 C.

2.2. Standards and reagentssugars and volatile compounds (Escriche, Visquert, Juan-Borrs,and Fito, 2009). However, the effect of industrial processing onthe antioxidant capacity of honey has not been studied in depth.A-mongst the few works dealing with this topic, the work of Turk-men, Sari, Poyrazoglu, and Velioglu (2006) should be mentioned.They observed that heat treatment increases the antioxidantcapacity of honey, resulting in a positive effect on human healthas a consequence of these Maillard action products, but also a neg-ative one, since browning caused by heating is not desirable byconsumers. Concerning this issue, it is important to highlight thepaper of Wang, Gheldof, and Engeseth (2004) who observed thatthe impact of traditional processing on honey antioxidant capacity(determined by oxygen radical absorbance) varies depending onthe type of honey. Buckwheat honey was more affected by process-ing than clover honey in terms of reduction in antioxidant capacity.In the same way, these authors showed that the impact of heatprocessing on the phenolic prole was complex; some compoundslike quercetin and galangin only increased signicantly in cloverhoney.

For the above mentioned reasons, the present work aims todetermine to what extent avonoids, phenolic compounds andthe antioxidant capacity of honey can be used in the authenticationof the botanical origin of four types of Spanish honey, and to eval-uate the impact of industrial thermal treatment on this process. Inorder to guarantee the reliability of the chromatographic proce-dure used to quantify the avonoids and phenolic compounds itwas validated before analysing the samples.

2. Materials and methodsHeat processing has been extensively evaluated considering the

136 I. Escriche et al. / Food ChAll the standards (purity higher than 99%): hesperetin, naringe-nin, caffeic acid, chrysin, p-coumaric acid, galangin, pinocembrin,kaempferol, myricetin and quercetin, were purchased from Extra-synthese (Lyon Nord 69726 France). Amberlite XAD-2 resin (poresize 9 nm and particle size 0.31.2 mm) was bought from Supelco(Bellefonte, PA, USA), ABTS [2,2-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt] was from Fluka Chemie, Buchs,Germany. Stock solution of ABTS (2 mM) was prepared by dissolv-ing in 50 mL of phosphate buffered saline (PBS), pH of the solutionwas adjusted to 7.4 with 0.1 M NaOH. The solvents used for thechromatography column were deionised water and analyticalgrade hydrochloric acid and methanol. For HPLC analysis, metha-nol and formic acid HPLC grade and bidestillated water were used,purchased from VWR (Darmstadt, Germany). The membrane lter(0.45 lm pore size) was acquired from Sartorius (Stedim Biotech,Germany).

2.3. Melissopalynological analysis

The analysis and quantication of pollen were carried out fol-lowing the recommendations of the International Commission forBee Botany (Von Der Ohe, Persano, Piana, Morlot, & Martin,2004). A light microscope (Zeiss Axio Imager, Gttingen, Germany)at a magnication power of 400 with DpxView LE image analysissoftware attached to a DeltaPix digital camera was used.

2.4. Physicochemical analysis

Physicochemical parameters [diastase activity, 5-hydroxymeth-ylfurfural content (HMF), moisture content and electrical conduc-tivity] were analysed following the recommendation of theHarmonised Methods of the European Honey CommissionBogdanov, (2002). Colour was determined using a millimetre Pfundscale and a spectrocolorimeter (Minolta CM-3600d, Osaka, Japan).Coordinates CIE Lab were obtained from R1 between 400 and700 nm for D65 illuminant and 2 observer. All tests were per-formed in triplicate.

2.5. Analysis of avonoids and phenolic acids

2.5.1. ExtractionThe extraction of avonoids and phenolic acids was carried out

as described by Baltruaityte et al. (2007): Sixty grams of amber-lite resin were soaked in methanol for 10 min, then, most of meth-anol was decanted and replaced by distilled water. The mixturewas stirred, allowed to stand for 510 min and was packed into aglass column (25 2 cm). Honey samples (25 g) were thoroughlymixed with 250 mL of distilled water and adjusted to pH 2 withconcentrated HCl. The solution was slowly ltered through the col-umn, which was washed with 250 mL of acidied water (pH 2 withHCl) and subsequently rinsed with 300 mL of neutral distilledwater to remove all sugars and other polar compounds of honey.The avonoids and phenolic compounds were eluted from the sor-bent with 250 mL of methanol. The extract obtained was concen-trated under vacuum at 40 C in a rotary evaporator (Laborota4000/HB/G1, Heidolph Instruments GmbH, Schwabach, Germany).The residue was dissolved in 5 mL of distilled water and extractedthree times with 5 mL of diethyl ether. The extracts were combinedand the solvent was removed by ushing with nitrogen. The driedresidue was then redissolved in 1 mL of methanol (HPLC grade)and ltered through a membrane lter. Three replicate extractionswere performed for each sample.

2.5.2. HPLC analysisAnalyses of the extracts were carried out by HPLC [Alliance

try 142 (2014) 1351432695, with a 2996 photodiode array detector (Waters, USA)]. Flavo-noids and phenolic compounds were separated on a Sunre TManalytical C18 column, 150 4.6 mm i.d. and particle size 5 lm

The percentage of pollen from the principal botanical species

stase activity and conductivity. Citrus and rosemary honeys had

emis(Waters, Ireland). The binary mobile phase consisted of solvent A(water and formic acid, 95:5) and solvent B (100% methanol).According to the method of Martos et al. (2000), the following gra-dient was used: starting with 30% methanol (B) which was keptisocratically for 15 min, and then followed by a gradient to obtain40% methanol at 20 min, 45% methanol at 30 min, 60% methanol at50 min, and 80% methanol at 52 min. Finally, isocratic elution with80% methanol was carried out until 60 min. The ow rate was1 mL/min and the injection volume was 30 lL. Chromatogramswere recorded at three wavelengths (290, 340 and 370 nm). Flavo-noids and phenolic acids were identied by comparison of chro-matographic retention times and UV spectral characteristics ofunknown analytes with authentic standards and available litera-ture data (Merken & Beecher, 2000). Calibration curves were con-structed via least squares linear regression analyses of the ratioof the peak area of each representative compound versus therespective concentration. Quantitative results were expressed asmg of compound per 100 g of honey.

2.5.3. Total antioxidant activity: ABTS+ radical cation decolourisationassay

The antioxidant activity of honey samples was determined bythe reaction with stable ABTS+ radical cation according to themethod described by Baltruaityte et al. (2007). ABTS+ was pro-duced by reacting 50 mL of ABTS stock solution with 200 lL of70 mM K2S2O8. The mixture was left to stand in the dark at roomtemperature for 1516 h before use. For the evaluation of antioxi-dant activity, the ABTS+ solution was diluted with PBS to obtain anabsorbance of 0.800 0.030 at 734 nm. Ten microlitres of honeyphenolic extract solution, obtained as was described in Section2.5.1., were mixed with 3 mL of ABTS+ solution. The absorbancewas read at room temperature after 1, 4, 6 and 10 min. PBS solutionwas used as a blank. The measurement was performed in triplicate.The percentage decrease of the absorbance at 734 nm was calcu-lated by the formula I = [(A-AA)/AB] 100; where: I = ABTS+ inhibi-tion %; AB = absorbance of a blank (t = 0 min); AA = absorbance of atested honey extract solution at the end of the reaction(t = 10 min).

2.5.4. Validation of the analytical method (HPLCDAD) for avonoidsand phenolic acids

Due to the fact that honey has a complex matrix, it is necessaryto validate the methods used to analyse its components. This fact isespecially important for those compounds that have very low con-centration levels such as avonoids and phenolic acids, in order toensure suitable levels of recovery and repeatability.

The validation parameters: linearity, accuracy and precision(repeatability: intraday precision and reproducibility: interdayprecision), were obtained according to the criteria described by theCommission Decision (2002) and by ICH (1996). The sensitivity ofthe method was proved with the limit of detection (LOD) and thelimit of quantication (LOQ) of the diferents analytes.

Linearity was evaluated using a matrix similar to honey (com-mercial glucose syrup, Roquette, Espaa, S.A.) as a blank. Thiswas fortied at 4 nal concentration levels (0.1, 0.2, 0.6 and1.2 mg/100 g) and submitted to the same extraction procedure asthe honey (n = 4). Linear calibration curves were constructed fromthe peak areas of the compounds versus the correspondent concen-trations of the fortied blanks. The accuracy of each avonoid andphenolic acid was evaluated with the recovery test using the samelevel concentrations as linearity (0.1, 0.2, 0.6 and 1.2 mg/100 g),(n = 3).

The precision of the method was expressed as the relative stan-

I. Escriche et al. / Food Chdard deviation (RSD), in this way the repeatability was calculatedfrom the analysis of six blank samples fortied at each of the fourspecied levels of fortication, and performed by the same opera-considerably lower diastase activity average values (8.53 and11.74) and electrical conductivity (0.17 and 0.22) than the poly-oral and honeydew types, with values of 19.74 and 20.54 for dia-stase and 0.37 and 1.00 for conductivity. Other authors haveshown that amongst oral honeys, citrus and rosemary types haverelatively low conductivity levels (Corbella & Cozzolino, 2006; Ter-rab, Gonzalez, Diez, & Heredia, 2003). As expected, polyoral hon-eys had intermediate values, given that they come from manyvaried notably according to the origin of the honey. For citrus hon-ey this percentage was between 15% and 37% and for rosemaryhoney it ranged between 15% and 31%. In the case of honeydewhoney, dew elements between 1.9% and 2.5% were found. The hon-eys were considered to be polyoral, if the percentage of a specictype of pollen was not enough for classication as monooral.

Table 2 shows the average values resulting from the multifactoranalysis of variance (ANOVA), including the F-ratio, obtained forthe physicochemical parameters (diastase activity, HMF, moisturecontent, electrical conductivity, colour Pfund, CIELab coordi-nates) and TAA (Total Antioxidant Activity).

Diastase activity and electrical conductivity were very differentdepending on the botanical origin of honey as previously shown byother authors (Escriche et al., 2011). All samples fullled the min-imum required level established by the European Directive (Coun-cil Directive, 2002), as in all cases diastase activity was well over800. As expected, honeydew honey had the highest levels of dia-tor on the same day. To evaluate the reproducibility the analyseswere repeated on three consecutive days. LODs and LOQs wereestimated by fortifying blanks at different concentrations andapplying the extraction procedure, and were determined as theamount of analyte for which signal-to-noise ratios (S/N) were high-er than 3 and 10, respectively. The stock standard solution of the allthe compounds was prepared by dissolving the appropriateamounts in methanol. This stock solution was stored at 20 C.

2.6. Statistical analysis

A multifactor analysis of variance (ANOVA) (using StatgraphicsPlus version 5.1.) was carried out to study the inuence of the typeof honey, the thermal treatments applied and the year of harvest-ing on the physicochemical parameters and on the avonoids andphenolic compounds of honey. Three factors were taken into con-sideration: the type of honey (citrus, rosemary, polyoral and hon-eydew), the thermal treatment applied (liquefaction/pasteurisation) and the year (2008/2009). The double interactionsbetween these factors were also considered. The method used formultiple comparisons was the LSD test (least signicant difference)with a signicance level a = 0.05. In addition to this, the avonoidsand phenolic compounds data were analysed using a PrincipalComponent Analysis (PCA) applying the software UnscramblerX.10.

3. Results and discussion

3.1. Melissopalynological and physicochemical characterisation ofhoney

The samples used in this study were accurately classied asbelonging to a specic botanical variety according to the melissop-alynological characterisation, the most reliable and useful way ofdetermining the botanical origin of honey.

try 142 (2014) 135143 137different types of nectar (Corbella & Cozzolino, 2006). The HMFcontent of raw samples was always quite low in this study, theonly exceptions being some rosemary samples with values up to

sed

nuses, a

p.,ther

7%

8%a, a

lianoth

is spnogy

satid ot

ulga

emis15.6 mg/kg, which demonstrates that almost all the raw honey wasfresh. The values of HMF increased a little with the heat treat-ments, reaching 7.65 mg/kg average value in liqueed samples

Table 1Percentage of pollen for the most abundant botanical species in the four batches analyorder to supplement the pollinic information.

Batch1 Batch2

Citrus(2008)

21% Citrus sp., 2% Rosmarinusofcinalis, Genista sp., Diplotaxiserucoides, and others

17% Citrus sp., 11% Rosmariofcinalis, Diplotaxis erucoidothers

Citrus(2009)

22% Citrus sp., 15% Prunus dulcis,Ceratonia siliqua, and others

18% Citrus sp., Taraxacum sQuercus sp. Labiadas, and o

Rosemary(2008)

25% Rosmarinus ofcinalis, 12%Diplotaxis sp., and others

15% Rosmarinus ofcinalis, 1Diplotaxis sp., and others

Rosemary(2009)

30% Rosmarinus ofcinalis, Genista sp.,Erica sp., Prunus dulcis, and others

23% Rosmarinus ofcinalis, 1Thymus sp., Satureja montanothers

Polyoral(2008)

23% Echium sp., 4% Helianthus annuus,Lavandula stoechas, and others

39% Echium vulgare, 16% Heannuus, 11% Lavandula, and

Polyoral(2009)

20% Anthyllis cytisoides, 14%Helianthus annuus, 15% Umbelliferae,and others

27% Rubus sp., 19% DiplotaxOlea europea, Crataegus moand others

Honeydew(2008)

HDE = 1.955% Castanea sativa, 11% Rubus sp., 4%Lavandula stoechas, Erica sp., andothers

HDE = 2.027% Erica sp, 18% Castanea10% Lavandula stoechas, an

Conductivity = 1.150 mS/cm Conductivity = 0.892 mS/cmHoneydew

(2009)HDE = 1.916% Rubus sp., 6% Thymus sp.,Cruciferae, Onobrychis viciifolia, andothers

HDE = 1.026% Rubus sp., 7% Calluna v11% Erica sp., and others

Conductivity = 0.893 mS/cm Conductivity = 1.390 mS/cm

138 I. Escriche et al. / Food Chand 8.18 mg/kg in pasteurised samples. In the same way, and asa consequence of the thermal treatment, the average value of dia-stase activity decreased slightly from 16.07 (ID) in the raw samplesto 13.80 in the pasteurised ones. This logical behaviour is related tothe implication of both parameters as indicators of freshness.

Moisture content in all samples met the requirements of theCouncil Directive of 2002 (maximum limited to 20 g/100 g in orderto avoid yeast fermentation), as the maximum value did not exceed17 g/100 g.

The higher the F-ratio (quotient between variability due to theconsidered effect and the residual variance), the greater the effectthat a factor has on a variable. According to this, conductivity, vari-ables related to colour (colour Pfund, and CIELab coordinates),HMF diastase, and ATT were most affected by the type of honey;whereas moisture was most affected by treatment. Honeydewhoney presented the greatest values of both antioxidant capacitiesand several physicochemical parameters when compared to nectarhoneys. These results are in accordance with those reported byVela, De Lorenzo, and Perez (2007) for Spanish honeys.

The factor year has an inuence on moisture and diastaseactivity but above all on conductivity. The highest number of sig-nicant interactions was found for the combination of factors hon-ey and year. This signicant interaction indicates that the value ofthese parameters (conductivity, moisture, Pfund colour scale, TAAand diastase) is different from year to year depending on the typeof honey. In the case of ATT the interaction between type of honeyand treatment was also signicant. Wang et al. (2004) observedthat the impact of traditional thermal processing on honey antiox-idant capacity varies depending on the type of honey, due to thecomplicated chemical composition (pigments, phenolic com-pounds, enzymes, Maillard reactions and other minor compounds)which varies between honeys from different oral sources(Baltruaityte, Rimantas-Venskutonis, and Ceksteryte, 2007;Saxena, Gautam, & Sharma, 2010). In this work, thermal treatmentcaused a signicant increase of ATT (from 59.34% inhibition in theraw sample to 63.18% in the pasteurised) as expected, as the lique-

for 2008 and 2009 for each type of honey. For honeydew, conductivity is also given in

Batch3 Batch4

nd15% Citrus sp., 8% Diplotaxiserucoides, Prunus dulcis, Vitisvinifera, and others

19% Citrus sp., Labiatae, Erica sp.,Salix sp., Hypecoum imberbe, andothers

s37% Citrus sp., Taraxacum sp.,Genista sp., Labiadas, and others

28% Citrus sp., type Taraxacum,Helianthus annuus, Erica sp., andothers

21% Rosmarinus ofcinalis, 17%Prunus sp., Thymus sp., Genista sp.,and others

19% Rosmarinus ofcinalis, 12%Hypecoum imberbe, Thymus sp., andothers

nd31% Rosmarinus ofcinalis, 8%Diplotaxis sp., Thymus sp., andothers

25% Rosmarinus ofcinalis, 21%Prunus dulcis, Diplotaxis sp., andothers

thusers

26% Echium vulgare, 21%Eucalyptus sp., 10% Lavandula, andothers

31% Echium sp., 14% Eucalyptus sp.,8% Helianthus annuus, and others

.,ma,

19% Anthyllis cytisoides, 18%Echium vulgare, 9% Prunus dulcis,and others

24% Salix sp.,16% Rosaceae, 10%Rosmarinus ofcinalis, and others

va,hers

HDE = 2.531% Echium sp., 14% Eucalyptus sp.,8% Helianthus annuus, and others

HDE = 2.343% Echium sp., 4% Helianthusannuus, Lavandula stoechas, andothers

Conductivity = 0.884 mS/cm Conductivity = 0.900 mS/cm

ris,HDE = 2.221% Rosaceae, 8% Thymus sp.,Quercus sp., Crucferae, Salix sp.,and others

HDE = 2.09% Thymus sp. 9% Rosaceae, Viciafaba, Helianthus annuus, Rubus sp.,and others

Conductivity = 0.891 mS/cm Conductivity = 0.896 mS/cm

try 142 (2014) 135143faction and pasteurisation temperatures applied are not very high(Turkmen et al., 2006). In the same way, the difference betweenyears for TAA was signicant. However, it is important to empha-sise that although treatment and year were signicant, the factorbotanical origin had the greatest inuence on the antioxidantactivity.

As other authors suggest, the antioxidant activity of differentkinds of honeys can be related to specic physicochemical param-eters and colour (Baltruaityte et al., 2007; Gonzalez-Miret, Terrab,Hernanz, Fernandez-Recamales, & Heredia, 2005). The possiblecorrelations between all of these parameters were checked in thiswork using Pearson correlation coefcients (r). Table 3 shows thecorrelation matrix obtained for each pair of variables. The numberin brackets is the P-value which tests the statistical signicance ofthe estimated correlations at the 95.0% condence level. As ex-pected, there was no correlation between HMF or moisture andthe rest of the analysed parameters. However, Aljadi & Kamaruddin(2004) show that Malaysian oral honeys with higher water con-tent had higher antioxidant capacity. Good correlations were ob-served in this work between all the parameters related to colour(especially with regard to Pfund scale), and conductivity, diastaseactivity and TAA. The correlation between colour and conductivity,colour and ID is widely accepted (Perez, Iglesias, Pueyo, Gonzalez,& De Lorenzo, 2007; Saxena et al., 2010). Different authors showedthat the colour intensity of honey is related to pigments such ascarotenoids and avonoids, and therefore, pigments have a rolein the antioxidant activities of the honey samples (Saxena et al.,2010). Estevinho, Pereira, Moreira, Dias, & Pereira (2008) demon-strated that dark honeys from Northeast Portugal have a signi-cantly higher level of antioxidant capacity than the clear ones.Escuredo, Silva, Valentao, Seijo, and Andrade (2012) observed thatcolour in Rubus honeys presented a signicant correlation with thetotal phenol content and some individual phenolic compounds.

Table2

Physico-ch

emical

parameters,

CIE

Lab

colour

and

TAA

(total

antiox

idan

tactivity)an

alysed

inho

ney

samples

(Citrus,

rosemary,

polyo

ralan

dho

neyd

ew)be

fore

and

afterhe

attreatm

ent(R

=raw;L=liq

uefaction

and

P=pa

steu

risation

)an

dye

ar(200

8an

d20

09),an

dANOVAF-ratioforeach

ofthethreefactors(hon

ey,treatmen

t,an

dye

ar)an

dtheirrespective

doub

leinteractions.

Hon

eyfactor

(H)

Treatm

entfactor

(T)

Yearfactor

(Y)

Factor

interactions

Physicochem

ical

parameters

Citrus

Rosem

ary

Polioral

Hon

ey-dew

ANOVA

F-ratio

RL

PANOVA

F-ratio

2008

2009

ANOVA

F-ratio

HT

TY

HY

Diastaseactivity

(ID)

8.53

c11

.74b

19.74a

20.54a

108*

**

16.07a

15.54a

13.80b

6*12

.42b

17.86a

92***

1.17

ns

3.46

ns

11**

HMF(m

g/kg

)4.48

c14

.71a

6.10

b3.91

c15

1***

6.08

b7.65

a8.18

a10

*9.10

a5.50

b78

***

1.72

ns

0.18

ns

2ns

Moisture

(g/100

g)15

.51b

16.11a

15.03c

15.31b

32***

16.67a

14.98b

14.97b

206*

**

15.09b

16.00a

133*

**

1.18

ns

171.79

***

22**

Con

ductivity(m

Scm

1)

0.17

d0.22

c0.37

b1.00

a11

075*

**

0.45

a0.45

a0.44

b5n

s0.33

b0.55

a37

974*

**

0.29

ns

0.48

ns

2352

8***

Pfundcolour(m

m)

15.50d

42.33c

77.83b

116.16

a14

10***

61.25a

63.37a

64.25a

2ns

61.00b

64.92a

11*

0.47

ns

2.56

ns

16**

CIELab

colour

L47

.99a

39.20b

34.32c

26.02d

576*

**

37.89a

36.41b

36.36b

7*36

.86a

36.90a

0.01

ns

2.46

ns

0.33

ns

3ns

a0.41

d5.15

c10

.53a

6.73

b13

4***

5.26

a6.00

a5.85

a2n

s5.79

a5.62

a0.21

ns

0.40

ns

0.14

ns

0.9n

s

b22

.46a

20.85a

b17

.27b

6.12

c46

***

16.09a

17.06a

16.88a

0.3n

s17

.03a

16.32a

0.42

ns

0.07

ns

0.14

ns

1ns

TAA(%

inhibition)

39.46c

57.87b

66.86b

80.42a

61***

59.34a

b56

.68b

63.18a

4**

63.27a

56.20b

12***

4.10

**

0.19

ns

14***

Foreach

factor,d

ifferentlettersin

each

row

indicate

sign

icantdifferen

cesat

95%conde

nce

levelas

obtained

bytheLSDtest.

ns=Not

sign

icant.

*p