Anal Bioanal Chem (2011) 400:2555–2563

DOI 10.1007/s00216-010-4614-7

ORIGINAL PAPER

Analysis of flavonoids in honey by HPLC coupled

with coulometric electrode array detection and electrospray
ionization mass spectrometry
Karoline Petrus & Heidi Schwartz & Gerhard Sontag

Received: 4 November 2010 / Revised: 12 December 2010 / Accepted: 15 December 2010 / Published online: 13 January 2011
# Springer-Verlag 2011

Abstract The analysis of flavonoids in unifloral honeys by to 3.4 mg/kg honey. Galangin, kaempferol, quercetin,
high-performance liquid chromatography (HPLC) coupled isorhamnetin, and luteolin were detected in all investigated
with coulometric electrode array detection (CEAD) is honeys, whereas hesperetin occurred only in lemon and
described. The compounds were extracted by a nonionic orange honeys and naringenin in lemon, orange, rhododen-
polymeric resin (Amberlite XAD-2) and then separated on dron, rosemary, and cherry blossom honeys.
a reversed phase column using gradient elution. Quercetin,
naringenin, hesperetin, luteolin, kaempferol, isorhamnetin, Keywords Honeys . Flavonoid profiles . HPLC-CEAD .
and galangin were detected in a coulometric electrode array HPLC-ESI-MS
detection system between +300 and +800 mV against
palladium reference electrodes, and their presence was
additionally confirmed by HPLC coupled with electrospray Introduction
ionization mass spectrometry. The method was applied to
analysis of 19 honeys of different varieties and origin. The A wide range of minor constituents like phenolic acids,
limits of detection and quantitation ranged between 1.6 and flavonoids, certain enzymes, carotenoid-like substances,
8.3 μg/kg and 3.9 and 27.4 μg/kg, respectively. The vitamins, organic acids, and Maillard products are present in
recoveries were above 96% in fluid and above 89% in honey. The composition is rather variable and primarily
creamy honeys. Some of these honeys (melon, pumpkin, depends on the floral source. Among other parameters,
cherry blossom, dandelion, maple, and pine tree honey) flavonoid as well as flavonoid glucoside profiles or the
were investigated for their flavonoid content and profile for presence of a single flavonoid can contribute to identify
the first time. Differences between honeys were observed unifloral honeys. A great number of such honeys are on the
both in flavonoid concentrations and in the flavonoid market and have to be checked regarding their quality,
profiles. The flavonoid concentrations ranged from 0.015 authenticity, and labeling in order to prevent consumer
deception. Therefore, there is a necessity for analytical
Published in the special issue Analytical Sciences in Austria with methods which enable reliable evaluation of unifloral honeys.
Guest Editors G. Allmaier, W. Buchberger, and K. Francesconi. So far, different methods have been developed for
K. Petrus : H. Schwartz : G. Sontag (*) determination of flavonoids in honey. Due to the complex
Institute for Analytical Chemistry, University of Vienna, matrix of honey and the low concentration of these
Währinger Strasse 38, compounds, numerous authors have proposed a sample
1090 Vienna, Austria
cleanup including a preconcentration step. The cleanup
e-mail: [Link]@[Link]
depends on the question whether flavonoid aglycones or
H. Schwartz glycosides should be analyzed. In the first case, the honey
Center for Analytical Chemistry, sample is dissolved in acidified water. Then, the flavonoid
Department of Agrobiotechnology (IFA Tulln),
aglycones can be extracted with ethylacetate [1, 2], by solid
University of Natural Resources and Life Sciences,
Gregor-Mendel-Straße 33, phase extraction on reversed phase materials [3–5] or on
1180 Vienna, Austria nonpolar resins with [6–13] or without [14–16] an
2556 K. Petrus et al.

additional extraction step with ethyl ether or 2-butanol [17] 14 times more sensitive than diode array detection. Inoue et
or further purification on a Sephadex column [18]. With the al. [43] investigated the radical scavenging activity caused
nonionic resins, Amberlite XAD-2 and XAD-4 sugars, by phenolic compounds with HPLC coupled with coulo-
acids, pigments, and disturbing compounds can be more metric electrode array detection (CEAD) without using the
effectively eliminated than on reversed-phase cartridges. high selectivity of this method for determination of single
Additionally, the aglycones, being less polar than glyco- phenolic compounds. Already in 1993, Jörg et al. [44] have
sides, are effectively bound on the resins. Only in some demonstrated that HPLC/CEAD is—due to its selectivity
papers no extraction step was applied, and the flavonoids and sensitivity—well suited for determination of phenolic
were determined by analyzing the compounds after disso- acid esters in different honeys with detection limits between
lution of the honey in pure or acidified water [19, 20]. As a 0.1 and 1 μg/kg.
consequence of the high concentration of polar and ionic The objective of the present study was to compare SPE
compounds, contamination of the analytical column has to extraction of flavonoids in honey with accelerated solvent
be expected if numerous honey samples are analyzed. For extraction (ASE) and to develop a method based on HPLC/
analysis of flavonoid glycosides, the sample was dissolved CEAD for qualitative and quantitative determination of
in pure water, and the glycosides were pre-concentrated on flavonoids in different honeys. Additionally, to verify the
a C18-SPE-cartridge and eluted with methanol [21, 22]. For presence of the flavonoids, these compounds should be
determination of glycosides, the extraction on reversed identified by HPLC/ESI-MS. Some not so well known
phase material is more appropriate than the extraction on unifloral honeys like those of melon, maple, pumpkin,
nonpolar resins. cherry blossom, dandelion, and one nonfloral honey (pine
Due to the polarity of the compounds, it is not surprising tree) should be investigated with this method.
that GC with flame ionization [17] or mass spectrometric
detection [23, 24] was applied only occasionally. Capillary
electrophoresis with UV [25–27] or mass spectrometric Materials and methods
detection [28] has the advantage of high separation
efficiency and shorter analysis time in comparison to Collection of honey samples
high-performance liquid chromatography (HPLC). Never-
theless, it was difficult to separate all flavonoids in one run, Honey samples (Table 1) of various floral sources were
and it was necessary to wash the capillary after each collected in different countries from 2006 to 2008.
analysis in order to attain a better reproducibility [25]. Still,
these methods are well suited for qualitative analysis of Chemicals and solutions
flavonoids in honey and propolis [25, 26, 28, 29].
The method of choice for qualitative and quantitative Galangin, hesperetin (purity ≥95%), isorhamnetin (purity ≥95%),
analysis of flavonoids is HPLC coupled with UV [1, 3, 6, 7, kaempferol (purity ≥96%), luteolin (purity ≥99%), myricetin
9–18, 30–38], electrochemical [20], or mass spectrometric (purity ≥96%), and quercetin dihydrate (purity ≥99%)
detectors [2, 4, 5, 19, 21, 22, 39–41]. Isocratic elution of were purchased from Sigma-Aldrich (Buchs, Switzerland).
phenolic compounds in honey extracts by HPLC/UV Methanol, HiPerSolvChromanorm gradient grade, was from
suffered from co-elution of compounds, and as a result of VWR (Vienna, Austria); ethyl acetate (Rotisolve® HPLC
the lack of selectivity, complex chromatograms were grade) and acetic acid (Rotipuran 100% p.a.) were from Carl
obtained. Therefore, gradient elution was proposed. Never- Roth (Karlsruhe, Germany). Ethanol (p.a.) was purchased
theless, UV detection is not sensitive enough to determine from AustrAlco (Öster. HandelsgesmbH, Spillern, Austria);
trace amounts of some flavonoids. Using ultra-performance hydrochloric acid (37% p.a.), sodium hydroxide (p.a.),
liquid chromatography coupled with time of flight mass potassium dihydrogenphosphate (p.a.), and ortho-phosphoric
spectrometry (UPLC-Q/TOF-MS), more phenolic com- acid (85% p.a.) were from Merck (Darmstadt, Germany); and
pounds could be identified compared to diode array Amberlite XAD-2 was obtained from Supelco (Vienna,
detection [2]. HPLC-MSn on an ion trap mass spectrometer Austria). Bi-distilled water was used in all experiments.
is a promising technique in the determination of the floral
origin of unifloral honeys, especially when the flavonoid Stock solutions Four milligrams of each flavonoid was
glycoside profiles should be investigated [21]. HPLC- dissolved in 10 ml of methanol by sonication. The solutions
electrospray ionization (ESI/MS) has proved as a valuable were stored in dark flasks at 4 °C.
method for qualitative and quantitative assay of some
flavonoids in propolis [42]. Standard mixture Fifty microliters of each stock solution
Liang et al. have shown [20] that electrochemical were pipetted into a 20-ml flask and filled up with
detection of phenolic compounds in citrus honey is six to methanol, yielding 1 mg/L of each compound. This
Analysis of flavonoids by HPLC/CEAD 2557

Table 1 Investigated honeys

Honey Color Consistence Year Provenience

Acacia Golden yellow, clear Liquid 2006 Burgenland, Austria

Buckwheat Golden brown Creamy 2006 Lower Austria, Austria
Maple Cream-colored Solid 2008 Lower Austria, Austria
Phacelia Reddish brown, clear Low viscosity 2006 Burgenland, Austria
Pumpkin Yellow Creamy 2006 Lower Austria, Austria
Raspberry Yellow Creamy 2008 Lower Austria, Austria
Orange (organic) Ivory-colored Creamy 2006 Italyb
Cherry blossom Red brown clear Liquid 2007 Cervia, Emilia Romagna, Italy
Dandelion Yellowish brown Creamy 2007 Cervia, Emilia Romagna, Italy
Melon Light yellow Creamy 2007 Cervia, Emilia Romagna, Italy
Rhododendron Yellowish Creamy 2007 Cervia, Emilia Romagna, Italy
Rosemary Dark yellowish Creamy 2007 Cervia, Emilia Romagna, Italy
Citrus blossom Golden yellow Creamy 2007 Zafferana Etnea, Sicily
Orange blossom Yellow Creamy 2007 Zafferana Etnea, Sicily
Lavender Golden brown, unclear Liquid 2007 Croatiab
a
Sage Brown Liquid Croatiab
a
Thyme Brown Liquid Crete, Greece
Pine tree Brown unclear Liquid 2007 Greeceb
Rape seed Dark brown Creamy 2007 Mecklenburg-Vorpommern, Germany
a
Date could not be confirmed
b
Geographical origin not known

solution was diluted with mobile phase A to the The solution was filtered over cotton wool, and 10 g of
corresponding concentrations and stored at 4 °C. preconditioned Amberlite XAD-2 resin were added to the
solution, and the resulting slurry was filled into a glass
Phosphate buffer (0.2 M) Potassium dihydrogenphosphate, column (25×1.5 cm). The column was washed with 40 ml
27.22 g, was dissolved in bi-distilled water, 2 ml of ortho- of acidified water (pH 2) and then with 60 ml of bi-distilled
phosphoric acid was added, and the solution was adjusted to water. The flavonoids were eluted with 50 ml methanol,
1 L with bi-distilled water. and the solvent was evaporated to dryness at 40 °C using
rotary evaporator. Then, the residue was ultra-sonicated in
1 ml of methanol and 1 ml of 0.02 M phosphate buffer, and
Pre-conditioning of Amberlite XAD-2 resin the solution was filtered through a 13-mm PTFE 0.45-μm
syringe filter (Alltech, Deerfield, IL, USA). This solution
The dry resin was suspended in methanol, and the suspension was injected into the HPLC/CEAD as well as into the
was stirred with a magnetic stirrer for 1 min. After 15 min, HPLC/ESI-MS system.
methanol was decanted and displaced by bi-distilled water
and stirred again for 1 min. Ten minutes before using the resin, Sample preparation based on accelerated solvent extrac-
water was removed, and acidified water (pH 2 adjusted with tion (ASE) Five grams of honey (orange, organic) unspiked
1 M hydrochloric acid) was added to the resin. or spiked with 50%, 100%, 150%, and 200% of the
estimated concentration of each flavonoid were mixed with
Sample preparation, equipment, and chromatographic 10 g diatomaceous earth (Mueller-Scherr, Linz, Austria)
conditions and placed into a 34-ml extraction cell adapted with a
cellulose filter. This cell was placed into a Dionex ASE 100
One hundred grams of each honey were homogenized by instrument (Dionex, Vienna, Austria). The flavonoids were
manual stirring and sonication. extracted with ethylacetate or ethanol/water (80:20, v:v)
under the following operating conditions: extraction tem-
Sample preparation with nonpolar resin Five grams of perature, 100 °C; extraction time, 5 min; flush volume,
honey were dissolved in 25 ml of acidified water (adjusted 50%; purge time, 90 s; and number of cycles, 2. The
to pH 2 with 1 M hydrochloric acid) under constant stirring. obtained extract was evaporated to dryness at 40 °C using
2558 K. Petrus et al.

rotary evaporator. The residue was dissolved in 1 ml of Fifteen microliters of the standard solutions or the
methanol. Then, 1 ml of 0.02 M phosphate buffer was sample extracts were injected into the HPLC/ESI-MS
added, and the solution was sonicated for 5 min. One system. Mobile phase C consisted of 0.5% acetic acid in
milliliter of the solution was filtered through a 13 mm methanol/water (20:80, v:v); mobile phase D contained
PTFE 0.45 μm filter, transferred into a vial, and this 0.5% acetic acid in methanol/water (80:20, v:v). Chromato-
solution was injected into the HPLC-CEAD as well as into graphic separation was achieved using gradient elution:
the HPLC/ESI-MS system. linear increase from 20% to 40% D within 30 min, isocratic
elution at 40% D from 30 to 55 min, linear increase to 60%
HPLC/CEAD The HPLC system consisted of a Merck D within 10 min, linear increase from 60% to 65% D from
Hitachi pump L-6200, an AS-2000A auto-sampler 65 to 75 min, isocratic elution at 65% D for 10 min,
(Merck Hitachi, Tokyo, Japan), a Phenomenex C18 pre- decrease to 20% D within 5 min, and re-equilibration at
column (4×3 mm; particle size, 5 μm; Phenomenex, 20% D from 90 to 110 min.
Torrance, CA, USA), a Nucleodur Sphinx RP analytical The flow was splitted (1:4) so that only a part of the
column (150×4.6 mm; particle size, 5 μm Macherey- eluate reached the ionization source. ESI was performed
Nagel, Düren, Germany), a thermostat (Merck Hitachi), with nitrogen as nebulizing gas (10 L/min), 40 psi
and a CEAD (ESA, Chelmsford, MA, USA) equipped nebulizing pressure, at 300 °C drying gas temperature.
with eight flow-through cells. The potentials of six The capillary voltage was set at −4.5 or +4.5 kV. The total
working electrodes were set at +300, +400, +500, ion chromatograms (TIC) in the range of 100 to 400m/z of
+650, +750, and +800 mV against palladium reference the standard solutions and the honey extracts were
electrodes. Due to the high background current, the last measured, and the extracted ion chromatograms (EIC) of
two cells were set at +100 and +200 mV. the deprotonated [M−H]− and protonated molecular ions
Fifteen microliters of the standard solutions or the [M+H]+ of the flavonoids (myricetin, 317.2 and 319.2;
sample extracts were injected into the HPLC/CEAD quercetin, 301.2 and 303.2; naringenin, 271.3 and 273.3;
system. Chromatographic separation was achieved at a luteolin, 285.2 and 287.2; hesperetin, 301.2 and 303.2;
flow rate of 1 ml/min at 20 °C using gradient elution. kaempferol, 285.2 and 287.2; isorhamnetin, 315.3 and
Mobile phase A consisted of methanol/0.02 M phosphate 317.3; and galangin, 269.2 and 271.2) were collected and
buffer (20:80, v:v) adjusted to pH 3.2 with 1 M sodium compared. Data were then integrated using the software
hydroxide; mobile phase B was made of methanol/0.02 M DataAnalysis™ version 3.3 (Bruker Daltonics).
phosphate buffer (80:20, v:v) adjusted to pH 3.2 with 1 M
sodium hydroxide. The following gradient was used:
linear increase from 0% to 40% B within 40 min, isocratic Determination of flavonoids
elution at 40% B from 40 to 65 min, linear increase from
40% to 70% B from 65 to 70 min, isocratic period at 70% Qualitative analysis The retention times and the peak areas
B from 70 to 85 min, linear increase to 80% B within of sample and standard compounds of the HPLC/CEAD
further 15 min, decrease to 0% B from 100 to 102 min, chromatograms were measured. The peak areas were
and re-equilibration at 0% B from 102 to 120 min. The plotted against the potentials of the working electrodes
software for data acquisition and evaluation was Coul and resulted in peak shaped current/voltage curves (Fig. 1).
Array Win (ESA). These voltammograms served as identification criterion for
flavonoids and for determination of the optimal detection
HPLC/ESI-MS The HPLC 1100 series system (Agilent potential (Table 2).
Technology, Palo Alto, CA, USA) consisted of a G1312A HPLC/ESI-MS measurements were performed to verify
binary pump, a G1322A mobile phase vacuum degassing the identification of the major peaks. Positive and negative
unit, a G1313A autosampler, a Nucleodur Sphinx RP ESI-MS spectra were recorded, and the extracted ion
analytical column (150×4.6 mm; particle size, 5 μm; chromatograms of the samples were compared with those
Machery-Nagel, Düren, Germany) protected by a Phenom- of the standard mixture.
enex C18, precolumn (4×3 mm; particle size, 5 μm;
Phenomenex, Torrance, CA, USA), and an ion trap mass Quantitative analysis The contents of the flavonoids in
spectrometric (MS) detector (HCT plus, Bruker Daltonics, honeys were determined by HPLC/CEAD at the optimal
Bremen, Germany) equipped with an ESI source. Hystar detection potentials on the basis of external calibration
3.1, software for chromatography and hyphenated techni- curves and using the standard addition method. Each
ques, was used for data acquisition. The compounds were sample was worked up in triplicate and injected once. For
separated at 20 °C at a flow rate of 1 ml/min using the quantitative analysis and recovery determination by stan-
gradient elution. dard addition method, the concentration of each flavonoid
Analysis of flavonoids by HPLC/CEAD 2559

Fig. 1 Current voltage curves

of galangin and hesperetin

in the unspiked sample was estimated by the external tenfold noise into the calibration curve of the corresponding
calibration curve, and then 50%, 100%, 150%, and 200% flavonoid. The intra-day reproducibility of the chromato-
of the experimentally determined concentration of each graphic method was evaluated by fivefold injection of the
flavonoid were added to aliquots (5 g) of the honey standard mixture containing 1 mg/L of each flavonoid in
samples. The spiked samples were processed as the 1 day and measuring the peak areas. The inter-day
unspiked. The analyte concentrations in the sample extracts reproducibility was assessed on the basis of five injections
were determined by plotting the peak areas versus the of the standard mixture containing 1 mg/L of each
added concentrations of the corresponding flavonoid, linear flavonoid over a period of 5 days. For determination of
regression, and division of the y-intersection (d) by the the inter- and intra-day reproducibility in sample solutions,
slope (k) of the regression line y=kx+d. one honey sample extract (orange, organic) was injected
Recoveries (N=5) were determined for a creamy (orange, five times in 1 day and five times each day during a period
organic) and a fluid honey (sage) by dividing the slope of the of 5 days. The limits of detection and quantitation in orange
standard addition regression line of each flavonoid by the (organic) and sage honey (Table 2) were determined by
slope of the external calibration function recorded during the inserting the three- and tenfold noise in the calibration
same analysis run and multiplying by 100. curve of the flavonoid.

Validation of the HPLC/CEAD method Results and discussion

Standard calibration curves of each flavonoid at the optimal Sample preparation

detection potentials were generated between 0.01 and
20 mg/L by plotting the peak area versus concentration of The extraction of the flavonoids with Amberlite XAD-2
the compound, and the correlation coefficients were resin used by Martos et al. [10] was modified in the
evaluated. The limits of detection in standard solutions following way. The amount of honey (100 g) for sample
were determined by dilution of the 0.01-mg/L standard preparation could be reduced due to the sensitivity and
until a signal to noise ratio of 3 was reached. The limit of selectivity of the HPLC/CEAD method. Only 5 g of the
quantitation (S/N=10) was determined by inserting the honey were extracted, and an additional extraction with

Table 2 Characteristic parameters of the HPLC/CEAD method

Compound Optimal detection Recovery of Recovery of Limit of detection Limit of quantitation CV (%) CV (%)
potential fluid honey (%) creamy honey (%) (S/N=3; μg/kg) (S/N=10; μg/kg) intraday interday
(channel; mV) (n=5) (n=5)

Myricetin +300 nd nd 8.2 22.3 nd nd

Quercetin +300 98.0±2.2 98.2±5.3 4.5 13.2 4.9 4.8
Naringenin +800 nd 93.1±7.8 4.7 27.4 5.8 9.6
Luteolin +300 100±6.1 89.2±7.07 4.5 13.1 4.0 6.1
Hesperetin +750 nd 95.4±9.5 3.7 8.9 4.0 8.4
Kaempferol +300 96.2±4.5 94.6±8.2 6.1 18.4 3.4 4.3
Isorhamnetin +300 99.6±3.5 89.9±4.1 8.3 21.7 4.6 9.9
Galangin +400 96.8±2.2 90.3±7.6 1.6 3.9 2.6 4.9

Creamy honey orange (organic), fluid honey lavender, nd not detected

2560 K. Petrus et al.

diethyl ether before injection as used for the HPLC/UV quercetin, naringenin, luteolin, hesperetin, kaempferol, iso-
method [10] was not necessary. The recoveries (N=5) of rhamnetin, and galangin could be identified. To the best of our
the flavonoids were in the range from 96% to 100% for knowledge, the flavonoid profiles of melon, pumpkin, cherry
fluid and between 89% and 98% for creamy honeys blossom, dandelion, maple, and pine tree honeys were
(Table 2). Additionally, ASE of the orange (organic) honey determined for the first time.
with different solvents (ethanol, methanol, ethyl acetate, As can be seen in Fig. 2, hesperetin is overlapped by
and water) was investigated. With ethyl acetate and ethanol/ another compound at low potentials (+300, +400,
water (80:20, v:v) as solvents recoveries between 82% to and +500 mV). Due to the coulometric detection mode, this
99% and 84% to 100%, respectively, were obtained. Based compound is fully oxidized at these potentials so that
on these data, ASE could be an alternative to extraction hesperetin could be quantified without any problem at the
with nonpolar resins, but it is more cost intensive. second maximum (+750 mV) of the current voltage curve.
So, with exception of myricetin (LOD, 8.2 μg/kg), all
Separation, detection of flavonoids, and validation investigated flavonoids could be quantified in at least some
of the method of the investigated honeys (Fig. 3). The flavonoid contents of
the unifloral honeys were different. In principle, one can
Due to the different polarity of the phenolic compounds in distinguish between honeys of low total flavonoid content
honeys, an isocratic elution with relatively low content of like citrus (0.66 to 1.77 mg/kg), lavender (0.91 mg/kg),
organic solvent, which would be optimal for electrochem- thyme (1.04 mg/kg), sage (1.15 mg/kg), acacia (1.6 mg/kg),
ical detection, is impossible. Hence, different gradient raspberry (2.28 mg/kg), and rhododendron (2.54 mg/kg)
programs were tested in order to achieve not only an honeys and those with a high content like cherry blossom
adequate separation of myricetin, quercetin, naringenin, (6.9 mg/kg), melon (5.78 mg/kg), pumpkin seed (4.93 mg/
luteolin, hesperetin, kaempferol, isorhamnetin, and galangin kg), rosemary (4.72 mg/kg), dandelion (4.35 mg/kg),
but also a good repeatability if a great number of samples phacelia (3.85 mg/kg), and buckwheat (3.26 mg/kg) honeys.
have to be investigated. Therefore, a relatively long While quercetin, luteolin, kaempferol, isorhamnetin, and
analysis method with appropriate gradient and washing galangin were present in all honeys, hesperetin was found
steps was developed. The current/voltage curves of the only in the three citrus honeys from Italy (continent and
flavonoid standards were established. This is shown in Sicily) in concentrations between 0.16 and 0.30 mg/kg. It
Fig. 1 for galangin and hesperetin. The optimum detection could be confirmed that this substance is characteristic for
potentials for determination of flavonoids are located in the those honeys [8, 12, 16, 26]. Hesperetin was accompanied
maxima of the current/voltage curves. At these potentials, by low concentrations (0.03–0.27 mg/kg) of naringenin, a
the calibration curves between 0.01 and 2 mg/L of the compound which has been observed commonly in citrus
investigated flavonoids were linear and showed correlation fruits and leaves. It is of interest that other authors [3, 10, 12,
coefficients (R2) between 0.9991 (quercetin) and 0.9997 16, 20, 26] could not identify naringenin in citrus and orange
(hesperetin). The standard addition curves had similar honeys. In this investigation, naringenin was also found in
correlation coefficients, e.g., 0.9995 for hesperetin in rhododendron, rosemary, and in the highest concentration
orange (organic) honey. The limits of detection and (0.27 mg/kg) in cherry blossom honey. The presence of
quantitation, the reproducibility of the method, and the naringenin together with kaempferol and galangin in honeys,
recoveries are summarized in Table 2. which were harvested in Cervia (Emilia Romagna, Italy) and
Zafferana Etnea (Sicily), could be explained by certain
Qualitative and quantitative analysis of honeys propolis amounts collected by bees from leaf buds and
cracks in the barks of lemon and orange trees. Volpi [29] has
Eighteen samples labeled as unifloral honeys (Table 1) and one analyzed phenolic compounds in propolis from the region
nonfloral honey (pine tree) were analyzed and evaluated. In Emilia Romagna and found—beside other phenolic com-
Fig. 2, the chromatogram of a creamy (orange, organic) honey pounds—naringenin, galangin, kaempferol, and quercetin,
is illustrated. For identification of the compounds, the shape of while luteolin and myricetin could not be detected. The
the current/voltage curves was compared with those of the highest concentration was observed for galangin and the
standard substances in addition to the retention time. relations between galangin/quercetin, galangin/kaempferol,
Additionally, HPLC/ESI-MS chromatograms in positive and and galangin/naringenin were about 2, 7, and 18, respec-
negative ionization mode yielding the characteristic protonated tively. In the honeys harvested in this region, we found
[M+H]+ and the deprotonated molecular ions [M−H]− were relations of 1.2, 3.5, and 15, respectively. Due to these data,
recorded. MS as independent detection mode served for it could not be confirmed that naringenin is a marker for
confirmation of CEAD data. The qualitative data of so far not lavender honey [1, 26]. We could not identify naringenin in
investigated honeys are presented in Table 3. In this way, lavender honey, either (Fig. 3).
Analysis of flavonoids by HPLC/CEAD 2561

Fig. 2 CEAD chromatogram

of an extract of orange (organic) 200
honey. Qu quercetin, Nar
naringenin, Lut luteolin,
Gal
Hes hesperetin, Kae kaempferol, Qu
Nar
Lut Hes
Irh isorhamnetin, Gal galangin 150
Irh
[800 mV] Kae

Response (nA)
[750 mV]
100
[650 mV]

[500 mV]

50
[400 mV]
6
5
[300 mV]
0 4
3
2
1
40.0 50.0 60.0 70.0 80.0
Time (min)

However, the absence of naringenin in dandelion and melon contents varied between 0.03 and 0.95 mg/kg for quercetin
honeys harvested in the same region (Cervia, Italy) by the same and 0.1 to 3.4 mg/kg for kaempferol. Therefore, these
apiculturist as rhododendron, rosemary, and cherry blossom substances could not be indicated as marker substances, even
honey (Table 2) shows that the interpretation of the data is though some honeys are characterized by a relatively high
difficult. Explanations might be that the concentration of content of kaempferol (melon, pumpkin, rape seed, and cherry
naringenin was below the detection limit or that another kind blossom honeys) in comparison to, e.g., herb honeys (Fig. 3).
of propolis had been transported into the hives by the bees.
The propolis composition, as shown by Volpi et al. [42], is
quite different and can influence the phenolic composition of Conclusion
honeys. On the other hand, melon honey from Cervia (Italy)
and pumpkin honey harvested in Lower Austria show very The floral source appears to be the primary reason for large
similar flavonoid profiles (Fig. 3) combined with the highest variations in the flavonoid content in honeys, but the
kaempferol contents of 3.4 and 2.8 mg/kg, respectively. This flavonoid content is also largely influenced by the kind and
could be a hint that these flavonoid profiles are characteristic content of propolis. The propolis composition depends—
for these kinds of honey and are not so much influenced by among other factors—on the geographical provenience, the
the propolis content. plants growing there, and the climate predominating in
Quercetin and kaempferol were proposed as markers for these regions [29, 42]. For an objective authentication of
sunflower and rosemary honey, respectively, by Tomás- the floral origins, the flavonoid profiles alone are not
Barberán et al. [12] and Gil et al. [9]. Both compounds were sufficient, but a number of other chemical parameters like
found in all kinds of honeys under investigation, and their metal ions [45] or volatile organic compounds [46] could

Table 3 Qualitative analysis of honeys by HPLC-ESI/MS

Positive mode (m/z) Myricetin Quercetin Naringenin Luteolin Hesperetin Kaempferol Isorhamnetin Galangin
319.2 303.2 273.3 287.2 303.2 287.2 317.3 271.2

Melon honey − + − + − + + +
Pumpkin honey − + − + − + + +
Cherry blossom honey − + + + − + + +
Dandelion honey − + − + − + + +
Maple honey − + − + − + + +
Pine tree honey − + − + − + + +

+ identified, − below the detection limit (0.8 to 4.6 μg/kg)

2562 K. Petrus et al.

Fig. 3 Flavonoid profiles 3.0

of honeys 2.5

c (mg/kg)
2.0
1.5
1.0
0.5
0.0

c (mg/kg)
3.0
2.0
1.0

0.0
0.8
c (mg/kg) 0.6
0.4
0.2
0.0
0.8
c (mg/kg)

0.6
0.4
0.2
0.0
0.3
c (mg/kg)

0.2
0.2
0.1
0.1
0.0
0.25
c (mg/kg)

0.20
0.15
0.10
0.00
0.35
c (mg/kg)

0.30
0.25
0.20
0.15
0.10
0.0
ia
ho Ac )
le

Pu on
ng icy

ap on
er

ck in
y

de a

Ph rry

rry
ge

d
on e

os on

at
ic

y
ne

el
ilic

do aci

ee
ym
ap

ar

Bu pk
nd

an

he
el
r
Sa

ac

li
e

he
Pi

Si

nd

em
de
S

es

pb
M

m
ve

Th

rg

C
e

an
as
(o
La

D
R

R
e
ra
Le

ng
O

ra

R
O

help to identify unifloral honeys. For instance, honeys with detection, a potential in the limiting current range has to be
low flavonoid contents like lavender, sage, and thyme chosen. This means that a potential of about +1,100 mV
honeys can be easily distinguished from honeys containing against silver/silverchloride/sat. potassium chloride would
relatively high flavonoid concentrations as, e.g., rosemary be necessary to detect the investigated flavonoids. At such a
honey. high detector potential, the recognition of overlapping
HPLC with coulometric electrode array detection proved substances in the chromatogram would be difficult. Coulo-
to be a reliable method for determination of traces of metric electrode array detection allows the quantification of
flavonoids in honey with detection limits in the low ppb substances simultaneously at lower potentials, and over-
range. This detection mode offers more selectivity and lapping of substances can be recognized if their electro-
detection of the compounds at lower potentials in compar- chemical behavior is different. The drawback of a long
ison to amperometric detection at one working electrode. analysis time for one sample was compensated by the
The optimal detection potential of a compound located in a stability of the total system. The only maintenance steps
peak maximum of the current voltage curve corresponds required over a period of 6 months (including analysis of
approximately to the half wave potential of a sigmoid 200 samples) were electrochemical cleaning steps by
hydrodynamic voltammogram established with a single cell polarizing the electrodes some minutes at −800 and then
electrochemical detector. In contrast, for single electrode by +900 mV in mobile phase A.
Analysis of flavonoids by HPLC/CEAD 2563

References 23. Berahia T, Cerrati C, Sabatier S, Amiot M (1993) Sciences des

Aliments 13:15–24
24. Sabatier S, Amiot MJ, Tacchini M, Aubert S (1992) J Food Sci
1. Amiot MJ, Aubert S, Gonnet M, Tacchini M (1989) Apidologie 57:773–774, 777
20:115–125 25. Delgado C, Tomás-Barberán FA, Talou T, Gaset A (1994)
2. Trautvetter S, Koelling-Speer I, Speer K (2009) Apidologie Chromatographia 38:71–78
40:140–150 26. Andrade P, Ferreres F, Gil MI, Tomás-Barberán FA (1997) Food
3. Pichichero E, Canuti L, Canini A (2009) J Sci Food Agric Chem 60:79–84
89:609–616 27. Gómez-Caravaca AM, Segura-Carretera A, Fernandez-Gutiérrez
4. Pulcini P, Allegrini F, Festuccia N (2006) Apiacta 41:21–27 A (2006) Agro Food Industry Hi-Tech 17:68–72
5. Michalkiewicz A, Biesaga M, Pyrzynska K (2008) J Chromatogr 28. Arraez-Román D, Gómez-Caravaca AM, Gómez-Romero M,
A 1187:18–24 Segura-Carretero A, Fernandez-Guitiérrez A (2006) J Pharmaceut
6. Gheldof N, Wang X-H, Engeseth NJ (2002) J Agric Food Chem Biomed Analysis 41:1648–1656
50:5870–5877 29. Volpi N (2004) Electrophoresis 25:1872–1878
7. Yao L, Jiang Y, D’Arcy B, Singanusong R, Datta N, Caffin N, 30. Kenjeric D, Mandic ML, Primorac L, Cacic F (2008) Food Chem
Raymont K (2004) J Agric Food Chem 52:210–214 110:187–192
8. Ferreres F, Garcia-Viguera C, Tomás-Lorente F, Tomás-Barberán 31. Ferreres F, Tomás-Barberán FA, Soler C, Garcia-Viguera C, Ortiz
FA (1993) J Sci Food Agric 61:121–123 A, Tomás-Lorente F (1994) Apidologie 25:21–30
9. Gil MI, Ferreres F, Ortiz A, Subra E, Tomás-Barberán FA (1995) J 32. Yao L, Datta FA, Tomás-Barberán FA, Ferreres F, Martos I,
Agric Food Chem 43:2833–2838 Singanusong R (2004) Food Chem 81:159–168
10. Martos I, Cossentini M, Ferreres F, Tomás-Barberán FA (1997) J 33. Yao L, Jiang Y, Singanusong R, D’Arcy B, Datta N, Caffin N,
Agric Food Chem 45:2824–2829 Raymont K (2004) Food Res Int 37:166–174
11. Martos I, Ferreres F, Yao L, D’Arcy B, Caffin N, Tomás-Barberán 34. Kenjeric D, Mandic ML, Primorac L, Bubalo D, Perl A (2007)
FA (2000) J Agric Food Chem 48:4744–4748 Food Chem 102:683–690
12. Tomás-Barberán FA, Martos I, Ferreres F, Radovic BS, Anklam E 35. Ferreres F, Juan T, Perez-Arquillue C, Herrera-Marteache A,
(2001) J Sci Food Agric 81:485–496 Garcia-Viguera C, Tomás-Barberán FA (1998) J Sci Food Agric
13. Iurlina MO, Saiz AI, Fritz R, Manrique GD (2009) Food Chem 77:506–510
115:1141–1149 36. Čeksterytė V, Kazlauskas S, Racys J (2006) Biologija 2:28–33
14. Weston RJ, Mitchell KR, Allen KL (1999) Food Chem 64:295– 37. Vit P, Tomás-Barberán FA (1998) Z Lebensm Unters Forsch A
301 206:288–293
15. Ferreres F, Tomás-Barberán FA, Gil MI, Tomás-Lorente F, Tomás- 38. Alabdeen Makawi SZ, Gadkariem EA, Ayoub SMH (2009) E-J
Lorente F (1991) J Sci Food Agric 56:49–56 Chem 6:S429–S437
16. Soler C, Gil MI, Garcia-Viguera C, Tomás-Barberán FA (1995) 39. Fiorani M, Accorsi A, Blasa M, Diamantini G, Piatti E (2006) J
Apidologie 26:53–60 Agric Food Chem 54:8328–8334
17. Weston RJ, Brocklebank LK, Lu Y (2000) Food Chem 70:427– 40. Martos I, Ferreres F, Tomás-Barberán FA (2000) J Agric Food
435 Chem 48:1498–1502
18. Vit P, Soler C, Tomás-Barberán FA (1997) Z Lebensm Unters 41. Baltrušaitytė V, Venskutonis PR, Čeksterytė V (2007) Food Chem
Forsch A 204:43–47 101:502–514
19. Biesaga M, Pyrzynska K (2009) J Chromatogr A 1216:6620– 42. Volpi N, Bergonzini G (2006) J Pharmaceut Biomed Analysis
6626 42:354–361
20. Liang Y, Cao W, Chen W-J, Xiao X-H, Zheng J-B (2009) Food 43. Inoue K, Murayama S, Seshimo F, Takeba K, Yoshimura Y,
Chem 114:1537–1541 Nakazawa H (2005) J Sci Food Agric 85:872–878
21. Truchado P, Ferreres F, Tomás-Barberán FA (2009) J Chromatogr 44. Joerg E, Sontag G (1993) J Chromatogr A 635:137–142
A 1216:7241–7248 45. Pohl P (2009) TRAC 28:117–128
22. Truchado P, Ferreres F, Bortolotti L, Sabatini AG, Tomás- 46. Cuevas-Glory LF, Pino JA, Santiago LS, Sauri-Duch E (2007)
Barberán FA (2008) J Agric Food Chem 56:8815–8824 Food Chem 103:1032–1043
Copyright of Analytical & Bioanalytical Chemistry is the property of Springer Science & Business Media B.V.
and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for individual use.