Volume 61, No. 4, 1996—JOURNAL OF FOOD SCIENCE—683 Flavor and Free Amino Acid Composition of Lavender and Eucalyptus Honeys AMINA BOUSETA, VINCENT SCHEIRMAN, and SONIA COLLIN ABSTRACT Unifloral honeys have highly characteristic flavors due to various vola- tiles probably derived from the original plant sources. The volatile com- pounds of honeys from two unifloral sources, lavender and eucalyptus, were investigated. Volatile compounds were extracted with dichloro- methane, followed by Likens-Nickerson’s simultaneous steam distilla- tion/solvent extraction. The extracts had an intense honey aroma that varied with the type of honey. Compounds (400) were separated by gas chromatography and detected by mass spectrometry. Major flavor com- pounds were very characteristic of each floral source (linear aldehydes, n-hexanol, coumarin, and phenylacetaldehyde in lavender honeys; di- ketones, hydroxyketones, 3-hexanol, sulfur compounds, and alkanes in eucalyptus samples). Key Words: honey, lavender, eucalyptus, amino acids, flavor INTRODUCTION FINDING RELIABLE MARKERS to ascertaining the floral origin of honey is a priority research objective in the apiculture industry. Floral origin is routinely authenticated by pollen analyses and physico-chemical and sensory determinations (Accorti et al., 1986; Apimondia, undated; Crane et al., 1984; Gonnet and Vache, 1984; Maurizio, 1979). Such data have been adequate for indexing several unifloral honeys (Barth, 1990; Feller-Del- masy et al., 1987a, b; 1989; Ricciardelli D’Albore, 1988). How- ever, the characteristic parameters available for certifying some floral origins are very limited. The authenticity of lavender (La- vandula augustifolia), citrus (Citrus ssp.), and rosemary (Ros- marinus officinalis) honeys, for example, is based solely on a minimum percentage of the specific pollen, the respective cri- teria being 10-13%, 15%, and 20% (Serra Bonvehi et al., 1987, 1988b). In some cases, when the plant is sterile (like hybrid lavender), sensory evaluations alone are used to identify floral origin. Unifloral honeys have highly specific aromas due to the pres- ence of specific volatile constituents probably derived from orig- inal nectar sources. The aim of our work was to study the aromatic content of honeys of two well-known unifloral origins, i.e.: lavender and eucalyptus. Chemical markers are especially needed for lavender honey, where the usual methods are inad- equate. We have investigated the headspace volatiles of honeys from 14 unifloral sources, including lavender and eucalyptus (Bouseta et al., 1992). Caproaldehyde (hexanal) and heptanal were detected in high amounts in lavender honeys only (up to 922 ppb and 241 ppb, respectively). On the other hand, 2,3- pentanedione, diacetyl, and dimethyldisulfide were found to- gether in eucalyptus honeys. Octane and nonane were also found in higher concentration in the eucalyptus samples. In the present work, our objective was to use an original op- timized extraction method (Bouseta and Collin, 1995) to inves- tigate the aromatic compositions of 10 lavender and 10 eucalyptus unifloral honeys, in comparison with 9 other honey types to identify the major flavor components as potential mark- Authors Bouseta, Scheirman, and Collin are affiliated with Univ- ersite ´ Catholique de Louvain, Unite ´ de Brasserie et des Industries Alimentaires, Place Croix du Sud 2/Bte 7, B-1348 Louvain-la- Neuve, Belgium. ers in distinguishing lavender and eucalyptus honeys from other types. Concentrations of some proposed flavoring markers were compared with probable precursors, i.e., phenylalanine and ty- rosine, determined by HPLC. MATERIALS & METHODS Honey samples Ten lavender and 10 eucalyptus unifloral honeys were selected from various countries (Lavender: samples 20, 22-24, 27-29, 31-32 from France; sample 25 from Spain. Eucalyptus: samples 10-11, 91-92, 94, 96-97 from Australia, 95 from Italy; samples 12 and 93 from Spain). Screening for floral purity was based on pollen analyses (Louveaux et al., 1978), sensory tests, conductivity, pH, titratable acidity (Journal Of- ficiel, 1977), and sugar composition (Pourtallier and Rognone, 1977). All samples regarded as unifloral honeys met requirements defined in other studies (Accorti et al., 1986; Apimondia, undated; Crane et al., 1984; Gonnet and Vache, 1984; Maurizio, 1979). The 90 samples of other unifloral origins (chestnut, fir, lime tree, orange blossom, rape, robinia, rosemary, sunflower and whiteclover, 10 of each) were also tested by the usual available physico-chemical and sensory tests. Reagents Acetoin (99%) and caproaldehyde (98%) were from Fluka Chemika (Buchs, Switzerland). Benzaldehyde (99+%), carvacrol (99%), coumarin, 2,3-dimethylphenol, 3,5-dimethylphenol (99%), heptadecane (99%), heptanal (95%), 3-hexanol (97%), isophorone (97%), 5-methylfurfural (99%), 2-methoxy-6-methylpyrazine (99%), pentadecane (99%), pheny- lacetaldehyde (95%), 2-phenylethanol (99%), and -pinene (98%) were from Aldrich Chemie (Steinheim, West-Germany). 2-Acetylfuran, ben- zyl alcohol (99+%), dimethyldisulfide (p.a), 2-furaldehyde (99%), fur- furyl alcohol (99%), hexanoic acid (99%), 3-methyl-2-buten-1-ol (99%), 3-methyl-3-buten-1-ol (97%), 2-methyl-1-butanol (98%), nonane (99%), octane (99+%), -terpinene (95%), and -valerolactone (98%) were from Janssen Chimica (Geel, Belgium). 3,4-Hexanedione and 1-phenylethanol (98%) were from Merck (Darmstadt, Germany). Dichloromethane (99.9%) was from Romil Chemicals (Leics, England). Propionic acid and butanoic acid were from Polysciences, Inc. (United Kingdom). Ben- zoic acid and -humulene were from Sigma Chimical (St-Louis, MO, USA). n-Hexanol (97%), phenylacetic acid, and toluene (99%) were from UCB Chemical (Leuven, Belgium). Phenol was from US Biochem- ical (Cleveland, Ohio, USA). 2-Methylbutanoic acid was from Alltech Associates, Inc. (IL, USA). Pyridine was from Analyticals Carlo Erba (Milano, Italy). Honey flavor extraction Solvent extraction was first performed to remove the sugar matrix which could induce artifacts due to nonenzymatic browning. After purg- ing the vessel with high-purity nitrogen, 100g of honey and 200 mL of twice distilled dichloromethane were poured into an extraction apparatus (Bouseta and Collin, 1995). The mixture was stirred for 60 min at 140 rpm under a 2 mL/min nitrogen stream to avoid oxidation reactions. The dichloromethane extract was concentrated to 1 mL in a Kuderna Snyder flask maintained in a 45°C water bath. Steam distillation-solvent extraction was carried out in a micro ex- tractor (Alltech 8910, composed of two flasks and a liquid/liquid ex- tractor) to remove flavor compounds from the co-extracted heavy matrix; this yielded an extract suitable for on-column chromatographic injection. The previously obtained 1-mL extract was transferred to the 100-mL flask with five 200-μL aliquots of dichloromethane used for washing the first extraction apparatus and 30 mL of ultrapure (Milli-Q water purifi- cation system, Millipore, Bedford, MA, USA), deoxygenated water.