Food Chemistry 79 (2002) 453–459

www.elsevier.com/locate/foodchem

Floral quality and discrimination of Lavandula stoechas,

Lavandula angustifolia, and Lavandula angustifolia
latifolia honeys
Christine Guyot-Declerck, Sarah Renson, Amina Bouseta, Sonia Collin*
Université catholique de Louvain, Unité de Brasserie et des Industries Alimentaires, Croix du Sud 2/Bte 7, B-1348 Louvain-la-Neuve, Belgium

Received 11 January 2002; received in revised form 12 April 2002; accepted 25 April 2002

Abstract
Portuguese lavender honeys are generated from the nectar of Lavandula stoechas, whereas French lavender honeys are exclusively
derived from Lavandula angustifolia, Lavandula latifolia, or hybrids of these two species. In the framework of the floral origin
authentication of such honeys, volatile compounds from L. stoechas, L. angustifolia, and L. angustifolia
latifolia unifloral honeys
were investigated. The aromatic profiles of French and Portuguese lavender honey samples showed major qualitative and quanti-
tative differences, but no volatile compound is characteristic of L. stoechas honeys only. As expected, n-hexanal, n-heptanal, phe-
nylacetaldehyde, and n-hexanol, previously proposed to authenticate French lavender honeys, were found at concentrations far
above the published discrimination thresholds. Coumarin, previously proposed to characterize French lavender honeys, emerges
here rather as an indicator of the freshness of lavender honey, being mainly released from glycosides during storage. Lastly, L.
angustifolia honeys were distinguishable from hybrid-derived samples by their lower phenylacetaldehyde and higher heptanoic acid
content. # 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Honey; Lavender; Flavour; Floral origin marker; Quality marker; Coumarin

1. Introduction It is hard to distinguish ‘‘fine lavender’’ from ‘‘lavan-

din’’ honeys. Melissopalynological characterization of
During the summer, Mediterranean landscapes are monofloral lavender honeys requires a percentage of
resplendent with the blue hues of lavender fields. These pollen ranging from 10 to 20% (Louveaux, Maurizio, &
plants of the genus Lavandula are cultivated or develop Vorwohl, 1978). These values are usually reached in L.
in a wild state. With their colours and odours peculiar to angustifolia samples, but never in L. angustifolia
latifolia
each species, the flowers of ‘‘fine lavender’’ (Lavandula honeys, because ‘‘lavandin’’ is sterile, and the number of
angustifolia), ‘‘spike lavender’’ (L. latifolia), ‘‘lavandin’’ pollen grains is especially low. For this reason, a threshold
(Lavandula angustifolia
latifolia) and ‘‘stechas lavender’’ of 50 grains per 10 g honey was proposed as a necessary
(Lavandula stoechas), constitute a prime nectar source (but not sufficient) condition for allowing the monofloral
for honey bees. Organoleptically, this specificity will be appellation (Loublier, Piana, Pham Delègue, & Borneck,
more or less pronounced in the honeys. 1994). Sensory analysis can be used as a complement, but
In France, lavender honey is protected by a ‘‘red requires experts and remains subjective.
label’’ (Gonnet, 1989). It derives exclusively from the In recent years, alternatives to sensory assessments
nectar of L. angustifolia, sometimes L. latifolia, or and pollen analyses (tedious and very dependent on
hybrids of these two species, to the exclusion of L. expert ability and judgment) have been developed in
stoechas (Journal Officiel de la République Française, order to characterize honeys more widely and accu-
1976). Its production area is essentially limited to the rately. L. stoechas honeys can be differentiated from
Southwest of France, L. stoechas honey being produced honeys from 10 other floral origins—heather (Erica sp.
in Portugal and Spain. and Calluna vulgaris), acacia, rape, sunflower, rosemary,
citrus, rhododendron, thyme and chestnut tree—on the
* Corresponding author. Tel.: +32-10-47-29-13; fax: +32-10-47-
basis of phenolic compounds, some of them specific like
21-78. naringenin, others predominant like m-coumaric acid
E-mail address: collin@inbr.ucl.ac.be (S. Collin). (Andrade, Ferreres, Gil, & Tomas-Barberan, 1997). High
0308-8146/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0308-8146(02)00216-9
454 C. Guyot-Declerck et al. / Food Chemistry 79 (2002) 453–459

contents in some aroma compounds, such as n-hexanal, Hungary, Russia and Spain; rosemary from France and
n-heptanal, n-heptanol, phenylacetaldehyde and cou- Spain; sunflower from Belgium and France; white clover
marin, are especially adequate for authenticating from Canada and New-Zealand; 10 of each) (Bouseta,
French lavender honeys among ten other floral origins 1994). Screening for floral purity was based on pollen
(Bouseta, Collin, & Dufour, 1992; Bouseta, Scheirman, analyses (Louveaux et al., 1978), sensory tests (Gonnet
& Collin, 1996). & Vache, 1984), conductivity, pH, titratable acidity
To our knowledge, no research has focused on estab- (Journal Officiel de la République Française, 1977), and
lishing distinctive floral markers of different species sugar composition (Pourtallier & Rognone, 1977).
within the genus Lavandula. Therefore, we have investi-
gated the volatile compounds of honeys derived from 2.1.2. Kinetics of the coumarin release and
three different species, L. stoechas, L. angustifolia, and determination of the coumarin potential
L. angustifolia
latifolia, using an optimised Likens- Two lavandin (L. angustifolia
latifolia) honeys were
Nickerson method yielding organoleptically highly used for this study. Each of them was divided into two
representative extracts (Bouseta & Collin, 1995). This halves stored at 4 and 40  C for 7 months.
technique recently proved adequate for distinguishing
C. vulgaris from Erica arborea heather honeys (Guyot, 2.2. Reagents
Scheirman, & Collin, 1999). In the present work, the
same methodology was applied to define markers for n-Hexanal (98%), n-heptanal (95%), n-octanal (99%),
each type of lavender honey. Amounts of coumarin n-nonanal (95%), benzaldehyde (90%), phenylacetalde-
were then compared in fresh and aged samples to assess hyde (90%), 5-methylfurfural (99%), furfurylalcohol
both the quality of the honeys and the reliability of this (99%), n-heptanol (98%), heptanoic acid (99%) and
previously proposed marker. coumarin were obtained from Aldrich (Bornem, Belgium),
3-methyl-2-buten-1-ol, 2-phenylethanol (99.5%), benzyl
alcohol (99%), hexanoic acid (99%), n-octane (99%), n-
2. Materials and methods nonane (99%) and 1-chloroheptane from Acros Chimica
(Geel, Belgium), n-hexanol (97%) and sulfuric acid
2.1. Honey samples from UCB (Bruxelles, Belgium), and 2-furaldehyde
from Acros (Geel, Belgium).
All the honeys were directly provided by the bee- The solvents (dichloromethane and methanol) were of
keepers and had not been industrially processed. They pure analytical grade (purity > 99.8%) and were pur-
were analysed just after harvest, unless otherwise stated. chased from Romil (Gent, Belgium). Dichloromethane
The samples were stored at 4  C until analysis. was redistilled twice prior to use. The water used was
ultra-pure water (Milli-Q water purification system,
2.1.1. Determination of the floral markers Millipore, Bedford, MA, USA). Sep-Pak classic short
Five L. stoechas from Portugal, six L. angustifolia
body C18 cartridges were obtained from Waters (Bruxelles,
latifolia (lavandin) and four L. angustifolia (fine laven- Belgium).
der) unifloral honeys from the Southwest of France
were used just after harvest in this study. Difficulties to 2.3. Honey flavour extraction
find ‘‘pure-species’’ samples explain why no more fresh
lavender honeys were analysed. Screening for floral Aroma compounds isolation was performed by a di-
purity was based on pollen analyses (Loublier et al., chloromethane dissolution, followed by a Likens–Nick-
1994; Louveaux et al., 1978; Pérez-Arquillué, Con- erson steam distillation/solvent extraction according to
chello, Ariño, Juan, & Herrera, 1995; Serra Bonvehi & the procedure described by Bouseta and Collin (1995).
Coll, 1993). Our data were also compared with ten For each sample, two replicates were obtained. The
French lavender honeys (without further genus char- reproducibility of the extraction method was previously
acterization) previously analysed by Bouseta et al. determined by Bouseta and Collin (1995) from five
(1996) after a 1-year storage at 4  C. consecutive analyses of a standard mixture. Depending
The aromatic profiles of Lavandula species honeys on volatile compound, coefficients of variation were
were here discriminated from those of 12 other unifloral found below 12% and recovery factors above 70%.
origins (C. vulgaris from France, Belgium, United
Kingdom, Norway and Germany; chestnut from France 2.4. Determination of the coumarin potential
and Italy; E. arborea from France, Greece and Italy;
Eucalyptus from Australia, Italy and Spain; fir from The protocol used is adapted from Abott (1991).
France; lavender from France; lime tree from France; Honey (25 g) was dissolved in 100 ml of ultra-pure
orange blossom from France, Mexico and Spain; rape water. The solution was then filtered through a Sep-Pak
from Belgium and France; robinia from Canada, France, C18 cartridge, previously activated by 50 ml of methanol
 C. Guyot-Declerck et al. / Food Chemistry 79 (2002) 453–459 455

and 50 ml of ultrapure water. The cartridge was rinsed rise from 36 to 85  C at 20  C/min then to 145  C at
with 50 ml of ultra-pure water in order to eliminate 1  C/min and to 250  C at 3  C/min. The carrier gas was
sugars and other water-soluble substances. The glyco- He at 1.5 ml min1. The injector temperature was
sides were then eluted using 50 ml of methanol. The maintained at 3  C above the oven temperature. The
methanol extract was evaporated to dryness in a rota- detector temperature was 260  C. The minimum peak
vapor (Heidolph, Germany). Fifty milliliters of H2SO4 area for data acquisition was set at 500 mV s. Retention
6N were added to the residue and the solution was indices were determined by interpolation of the reten-
placed in a water-bath for 16 h at 37  C  1  C and then tion times of a n-alkanes (C6–C19) mixture analysed
for 5 h at 60  1  C. The free coumarin isolation was under identical conditions. Two microliters of each
performed by three successive extractions with 50 ml of extract were injected for analysis.
dichloromethane. The dichloromethane extract was
concentrated to 1 ml in a Snyder Kuderna apparatus at 2.6. Quantification of volatile compounds by GC–FID
45 1  C with 50 mL of a 1000 mg g1 solution of 1-
chloroheptane in dichloromethane, added as external Concentration of compounds in the honey samples
standard. The extract was further analysed by GC. For was calculated with respect to the external standard,
each sample, two replicates were obtained. according to the equation:

2.5. Gas chromatography–FID (GC–FID) analytical Ci ¼ ðPi =Pe Þ
ðQe =Qh Þ
ð1=Ki Þ ð1Þ
conditions
where the suffix i, e and h refer to the quantified com-
A Hewlett Packard Model 5890 gas chromatograph pound, the external standard and the honey respec-
was used, equipped with a Hewlett Packard Model 7673 tively; P refers to the peak area obtained in GC–FID; C
automatic sampler, a cold on-column injector, a flame refers to the concentration in the honey (ng g1); Qe
ionization detector, and a Shimadzu CR4A integrator. refers to the quantity of external standard in the di-
Analysis of honey volatile compounds was carried out chloromethane extract (ng); Qh refers to the quantity of
on a 50 m
0.32 mm i.d. wall-coated open tubular honey used (g); Ki is the response factor at the FID
(WCOT) CP-SIL5 CB (Chrompack, Antwerp, Belgium) detector of the compound i with regard to the external
capillary column (film thickness, 1.2 mm), preceded by a standard.
1 m
0.53 mm i.d. capillary column, coated with a thin As previous analyses have shown that the Likens–
film of methyl silicone phase (Hewlett Packard, Brussels, Nickerson-derived method leads to recovery factors
Belgium). The oven temperature was programmed to higher than 70% for most of the chemicals mentioned

Table 1
Volatile compounds in Lavandula stoechas, Lavandula angustifolia and Lavandula angustifolia
latifolia honeysa

Compounds RI L. stoechas L. angustifolia L. angustifolia
latifolia

Min Max Avg Min Max Avg Min Max Avg

Pyridine 712 8 1130 242 0 0 0 0 0 0

3-Methyl-2-buten-1-ol 749 24 129 80 146 207 179 64 238 147
n-Hexanal 774 7 32 18 613 1460 939 980 1845 1346
Octane 800 9 27 15 10 57 27 17 37 26
2-Furaldehyde 803 160 223 192 53 154 95 65 182 102
Furfurylalcohol 824 6 21 12 17 48 32 0 54 18
n-Hexanol 844 0 0 0 1630 4370 2729 1886 4930 3983
n-Heptanal 877 0 73 33 179 329 286 185 294 238
n-Nonane 900 0 0 0 2 11 8 0 10 3
5-Methylfurfural 929 0 27 17 32 124 71 54 96 78
Benzaldehyde 933 5 194 74 31 91 62 82 151 111
Hexanoic acid 946 7 4388 1766 0 235 80 0 30 7
n-Heptanol 947 4 33 13 521 754 566 416 884 715
n-Octanal 979 4 9 7 13 56 33 41 75 61
Benzylalcohol 1009 9 113 65 21 44 32 31 101 57
Phenylacetaldehyde 1013 74 1329 703 744 1303 964 1539 2969 2189
Heptanoic acid 1049 5 30 19 193 296 238 85 194 132
n-Nonanal 1081 255 988 577 1135 1648 1427 1508 2163 1787
2-Phenylethanol 1087 130 2010 1132 730 1172 904 728 1242 971
Coumarin 1397 0 0 0 101 253 193 62 292 201
a
RI=retention index; Min, max, avg=minimal, maximal, average concentrations (ng g1) in the honeys analysed just after harvest.
456 C. Guyot-Declerck et al. / Food Chemistry 79 (2002) 453–459

(Bouseta & Collin, 1995), concentrations were calcu- peak enhancement by co-injection with authentic stan-
lated with an extraction recovery factor equal to dard compounds and comparison with the NBS/EPA/
100%. NIH mass spectra library.

2.7. Gas chromatography–mass spectrometry (GC–MS)

conditions 3. Results and discussion

Chromatographic conditions were the same as those 3.1. Distinguishing Portuguese and French lavender
used for FID detection. The column was directly con- honeys
nected to an HP 5988 quadrupole mass spectrometer.
Electron impact mass spectra were recorded at 70 eV After identification by GC–MS, 20 aroma compounds
(filament current: 300 mA; electron multiplier voltage: were quantified in Lavandula extracts by GC-FID
2500; scan rate: 4 s1; m/z range: 40–250). Spectral (Table 1). Most of these peaks do not constitute reliable
recording throughout elution was automatic using markers, due to their presence in honeys of other ori-
HP59970C software. Identification was on the basis of gins.

Fig. 1. Floral origin markers of French lavender honeys (Lavandula angustifolia and Lavandula angustifolia
latifolia): n-hexanal (A), n-heptanal
(B), n-hexanol (C), phenylacetaldehyde (D) and heptanoic acid (E). Honey analysed just after harvest ^; after storage at 4  C for 1 year or more
(Bouseta et al., 1996)* ~; not quantified
. * For the data derived from Bouseta et al. (1996), all French lavender honeys were analysed without
defining the floral species.
 C. Guyot-Declerck et al. / Food Chemistry 79 (2002) 453–459 457

The aromatic profiles of French (L. angustifolia and n-heptanal, hexanol, and phenylacetaldehyde are four
L. angustifolia
latifolia) and Portuguese (L. stoechas) typical markers of these lavender honeys (concentrations
lavender honey samples showed major qualitative and above 279, 61, 1594, and 744 ng g1, respectively)
quantitative differences. No marker specific to L. stoechas (Fig. 1A–D). From a qualitative point of view, the aro-
honeys was found (Figs. 1A–E and 2A–B), but among matic profiles of the L. angustifolia and L. angustifo-
honeys derived from flowers within the Lavandula lia
latifolia honeys proved very similar. However,
genus, Portuguese lavender honeys show much lower authentication between both types could be done on the
concentrations of n-hexanal, n-heptanal, n-hexanol and basis of their phenylacetaldehyde and heptanoic acid
heptanoic acid (below 32, 73, 0 and 30 ng g1, respec- contents (Figs. 1D and 1E): ‘‘lavandin’’ honeys exhibit
tively) than French lavender honey samples. Compared higher phenylacetaldehyde and lower heptanoic acid
to all other floral origins, these honeys are not easily concentrations (above 1539 and below 194 ng g1,
identifiable. Their authentication could be based on the respectively).
simultaneous absence of n-hexanol and n-nonane
(Figs. 1C and 2A) and the presence of pyridine (Fig. 2B), 3.3. Kinetics of coumarin release and determination of
though this last compound is almost negligible, except the coumarin potential
in one sample.
Results presented in Table 1 indicate lower coumarin
3.2. Authentifying the floral origin of L. angustifolia concentrations in our French lavender honeys analysed
and L. angustifolia
latifolia honeys just after harvest (from 62 to 292 ng g1) than in the 1
year-aged honey samples analysed by Bouseta et al. in
For L. angustifolia and L. angustifolia
latifolia 1996 (from 512 to 1720 ng g1, with an average equal to
honeys, our findings for honeys analysed just after harvest 954 ng g1). On this basis, we suggested progressive
confirm the finding of Bouseta et al. (1996) that n-hexanal, release of coumarin through storage, probably mainly
at the beginning due to residual enzymatic activity. This
phenomenon could be masked in longer storages due to

Fig. 2. Volatile compounds that distinguish Lavandula stoechas hon-

eys from honeys of other floral origins: n-nonane (A) and pyridine (B),
Honey analysed just after harvest ^; after storage at 4  C for 1 year or
more (Bouseta et al., 1996)* ~. * For the data derived from Bouseta
et al. (1996), all French lavender honeys were analysed without defin- Fig. 3. Coumarin metabolic pathway (Herbert, 1989; Murray et al.,
ing the floral species. 1982; Neish, 1965).
458 C. Guyot-Declerck et al. / Food Chemistry 79 (2002) 453–459

content in linear aldehydes, linear alcohols, and pheny-

lacetaldehyde. This last compound also emerges as a
quantitative marker of L. angustifolia
latifolia honeys,
whereas heptanoic acid preponderates in L. angustifolia
honeys.
Coumarin, which seems to be released from glycosides
through storage, could be used as a freshness marker of
French lavender honeys.

Acknowledgements
Fig. 4. Kinetics of coumarin release in ‘‘lavandin’’ honeys. ^ Sample
1;
sample 2; — at 4  C; -.-.-.- at 40  C; . . . . total coumarin content This work was supported by a grant from the Eur-
determined after acid hydrolysis. opean Commission, project FAIR-CT985050. We are
grateful to Paulo Nunes and Paulo Russo for providing
inevitable loss or hydrolysis of free coumarin, as sug- us with the Portuguese Lavandula stoechas honey sam-
gested by the analyses of six 4 year-aged samples (cou- ples and to Bernard Oza (SCA Provence Miel) and Jean-
marin concentration below 346 ng g1 in all cases). Louis Lautard for supplying us with French Lavandula
Coumarin is a shikimic pathway derivative resulting angustifolia and Lavandula angustifolia
latifolia honeys.
from phenylalanine metabolism (Fig. 3). Hydroxylation
of cinnamic acid in the ortho position coupled with glu-
cosylation is probably responsible for the formation of References
o-coumaric acid glucoside. This glucoside then under-
Abott, A. (1991). The study of Shiraz grape berry composition in rela-
goes a trans!cis conversion, possibly due to the action tion to the quality of table wine. PhD thesis, Adelaide University.
of sunlight, yielding coumarinic acid glucoside. Cleavage Andrade, P., Ferreres, F., Gil, M. I., & Tomas-Barberan, F. A. (1997).
of the glycoside bond by a specific b-glucosidase yields Determination of phenolic compounds in honeys with different
unstable cis-coumarinic acid, which spontaneously lac- floral origin by capillary zone electrophoresis. Food Chemistry,
60(1), 79–84.
tonizes to coumarin (Herbert, 1989; Murray, Mendez, &
Bouseta, A. (1994). Caracterisation et voies de formation des flaveurs du
Brown, 1982; Neish, 1965). Two lavandin honeys were miel [Characterization and formation pathways of honey flavours].
used for this study. Each was divided into two parts, PhD thesis, Catholic University of Louvain, Louvain-la-Neuve,
one being stored at 4  C and the other at 40  C for 7 Belgium.
months. The free and total coumarin contents are pre- Bouseta, A., & Collin, S. (1995). Optimized Likens-Nickerson metho-
sented in Fig. 4. dology for quantifying honey flavors. Journal of Agricultural and
Food Chemistry, 43(7), 1890–1897.
Storage at 40  C leads to rapid release of free cou- Bouseta, A., Collin, S., & Dufour, J.-P. (1992). Characteristic aroma
marin, a maximum being already reached after 3 profiles of unifloral honeys obtained with a dynamic headspace GC–
months. Release is as expected much slower at 4  C, MS system. Journal of Apicultural Research, 31(2), 96–109.
remaining significant even after 6 months. The free Bouseta, A., Scheirman, V., & Collin, S. (1996). Flavor and free amino
coumarin concentration reached the total coumarin acid composition of lavender and Eucalyptus honeys. Journal of
Food Science, 61(4), 683–687 694.
level obtained by acid hydrolysis (free+bound to the Gonnet, M. (1989). ‘‘Label Rouge’’ National pour un miel de crû.
glycoside form) in one honey sample, whatever the [Red label for a local honey]. L’Abeille de France, 741, 389–391.
temperature. Gonnet, M., & Vache, M. (1984). Le goût du miel. [Honey flavour].
These results suggest that coumarin can no longer be Paris, France: U.N.A.F. Editions.
Guyot, C., Scheirman, V., & Collin, S. (1999). Floral origin markers of
considered as a floral origin marker of French lavender
heather honeys: Calluna vulgaris and Erica arborea. Food Chemistry,
honeys, although it could be very useful for checking 64(2), 3–11.
the freshness of such samples. Herbert, R. B. (1989). The biosynthesis of secondary metabolites. Lon-
don: Chapman and Hall.
Journal Officiel de la République Française. (1976). Decree n 76–717,
4. Conclusion 22 July, Annexe 2, 37–46.
Journal Officiel de la République Française. (1977). Decree n 77–79,
22 April, related to the official analysis methods of honey, 3485–
In agreement with organoleptic observations (Gonnet 3514.
& Vache, 1984), our preliminary results evidence that Loublier, Y., Piana, M. L., Pham Delègue, M.-H., & Borneck, R.
honeys derived from different lavender species (L. stoe- (1994). Caractérisation pollinique des miels français de lavande:
chas, L. angustifolia and L. angustifolia
latifolia) show premiers résultats. [Characterization of monofloral lavender hon-
eys—pollen analysis of French lavender honeys—preliminary
specific volatile compounds profiles. French lavender results]. Grana, 33, 231–238.
honeys can be easily authenticated from L. stoechas Louveaux, J., Maurizio, A., & Vorwohl, G. (1978). Methods of
samples or various other origins, thanks to their high melissopalynology. Bee World, 59, 139–157.
 C. Guyot-Declerck et al. / Food Chemistry 79 (2002) 453–459 459

Murray, R. D. H., Mendez, J., & Brown, S. A. (1982). The natural Pourtallier, J., Rognone, C. (1977). Méthode améliorée de dosage des
coumarins. Occurrence, chemistry and biochemistry. Chichester: John sucres des miels par chromatographie en phase gazeuse [Improved
Wiley & Sons. method of sugars measurement in honeys by means of gas chroma-
Neish, A. C. (1965). Coumarins, phenylpropanes, and lignin. In tography]. In Proceedings of the Symposium International de Tech-
J. Bonner, & J. E. Varner (Eds.), Plant biochemistry (pp. 581–617). nologie Apicole (pp. 73–83). Bucarest: Apimondia.
New-York: Academic Press. Serra Bonvehi, J., & Coll, F. V. (1993). Physico-chemical properties,
Pérez-Arquillué, C., Conchello, P., Ariño, A., Juan, T., & Herrera, A. composition and pollen spectrum of french lavender (Lavandula
(1995). Physicochemical attributes and pollen spectrum of some stoechas L.) honey produced in Spain. Zum Lebensmittel Untersu-
unifloral Spanish honeys. Food Chemistry, 54, 167–172. chung und Forschung, 196, 511–517.