Chemical Variability of Essential Oils from Corsican Hops and Highlighting Their Influence on Hops’ Aroma

Essential oils from wild Corsican hops have never been described before. Following selective harvesting and extraction of plant material, chemical analysis was performed by GC–FID and GC–MS. Subsequent quantitative analysis demonstrated significant inter-stations variability due to pedoclimatic conditions. These singularities produce organoleptic differences, especially within Italian hops, which are the current benchmark for the Mediterranean hops’ population. Corsican wild hops are no exception. Accordingly, three olfactive bouquets were identified by a panel of selected and trained sensory analysts: woody herbaceous ginger notes, herbaceous citrus notes, and common notes. These bouquets appeared to be correlated to pedoclimatic parameters mentioned earlier such as altitude and proximity to the sea. A very rare and appreciated bouquet was associated with high levels of zingiberene in hops growing at moderate altitude and relatively far from the coastline. This study shows the importance of growing sites and pedoclimatic conditions to produce hops with the desired organoleptic notes during the beer making process and provides detailed identification of essential oils from Corsican wild hops.


Introduction
Hops (Humulus lupulus L.) are one of the four components of beer. Since the 13th century, the inflorescences of these vines, called cones, have been brewed for their aroma, bitterness, and preservative properties [1]. Lupulin, a green substance stored between the bracts of the strobili, is known to contain the compounds responsible for the defensive, curative, and organoleptic characteristics of hops [1][2][3]. The chemical composition of hops' essential oils has been well studied in the brewing industry due to their impact on the aroma profile of beers. Worldwide, several studies focused on their chemical characterization. The most common method used to identify and quantify volatile components is gas chromatography coupled with mass spectrometry, which is used in this study. The three main components identified in hops' essential oils are usually myrcene, humulene, and caryophyllene [4][5][6][7][8][9]. These molecules and their abundance are, therefore, responsible for the development of the desired beer aroma during the process. However, the fact that components usually found in minor concentration can also contribute to the development of unusual aroma profiles (bouquets) cannot be excluded.
Hops grow spontaneously between the 35 • and 55 • parallel north [10]. In Corsica (43 • N), wild hops are naturally present on riverbanks and mouths as well as on northern sandy soils. Even if the first mention of wild hops growing in Corsica dates back more than a century [2], their chemical profiles were only first described in one of our recently published works [11]. Moreover, the geographical variability of essential oils from spontaneous Corsican hops was evaluated for the first time in this study. It is crucial to point out that differences in essential oil content were reported for Italian hops growing over short distances due to the particularity of the Mediterranean conformation [12].

Plant Material
In 2019, after a 1500 km prospection around the island, cones of spontaneously growing hops were manually harvested at maturity in thirteen remote locations on Corsican riverbanks. At each station, samples were collected from three different individuals to limit possible bias due to theoretical inter individuals' variations. This sampling protocol was repeated in 2020 and 2021 in these same locations, whose soil characteristics are indicated in Table 1. A total of 117 samples were collected. Figure A2, found in Appendix A, provides a detailed map of the 13 locations across the island.

Essential Oils
Essential oils from the 117 samples were prepared by hydro-distillation using a Clevenger apparatus for four hours as recommended by the European Pharmacopoeia [17], then analyzed by GC-FID and GC-MS using the following devices.
Analyzes were carried out using a PerkinElmer Autosystem GC apparatus (Whaltam, MA, USA) equipped with a dual-flame ionization detection (FID) and fused silica capillary columns (60 m × 0.22 mm i.d., film thickness 0.25 µm) with different stationary phases: Rtx-1 (polydimethylsiloxane) and Rtx-Wax (polyethylene glycol). The oven temperature program was from 60 to 230 • C at 2 • C/min and then was held isothermally (30 min). Carrier gas was hydrogen (1 mL/min). Injector and detector temperatures were held at 280 • C. Split injection was performed with a fractional ratio of 1:80. The injected oil volume was 0.2 µL. The relative percentage of the oil constituents was calculated from the GC peak areas without using correction factors.
Mass spectrometry analyzes were conducted on a PerkinElmer TurboMass (Waltham, MA, USA) detector (quadrupole), coupled in line to a PerkinElmer Autosystem XL equipped with two fused silica capillary columns (60 m × 0.22 mm i.d., film thickness 0.25 µm), Rtx-1 (polydimethylsiloxane), and Rtx-Wax (polyethylene glycol). The other GC conditions were the same as described above. The ion source temperature was 150 • C, the energy ionization was 70 eV, and the electron ionization (EI) mass spectra were acquired over the mass range of 35−350 Da. The volume of oil injected was 0.2 µL. The main volatile compounds were identified based on their mass spectra compared with the reference mass spectra libraries (WILEY275, Wiley Science Solutions, USA; NBS75K, NIST, USA) and of their calculated retention indexes (RIs) through the application of the Kovats index (KI) formula compared with those reported in the literature and linear retention index (LRI). When it was not possible to find the KI in the literature, a tentative identification was obtained by matching with mass spectra libraries data: a match quality of 98% minimum was used as a criterion. The gas chromatographic signals were manually integrated, and the resulting peak areas were compared with the total sum of area and expressed in percentage.
For each year, the quantitative characterizations of the three samples collected at each location were averaged because of their similarity. The 39 chemical profiles obtained, and their corresponding harvesting year, are listed in Table 2.

Sensory Analyzes
A panel of twenty volunteers was trained to carry out sensory analyzes for several months according to the ISO 8586:2012 and 8588:2017 norms. These norms explain the selection process for effective panelists.
As requested, thirty-nine "naïve assessors" were confronted with simple blind sensory analyzes consisting of determining the presence or absence of a product with a marked fragrance (e.g., cinnamon) in a hydro-ethanolic solution contained in an opaque bottle to avoid visual bias. The twenty-eight sensitive enough panelists became "initiated assessors". They were trained to recognize the five typical scent categories of hops ("spicy", "woody", "herbal", "citrus", and "tropical fruit") and asked to determine their intensity from 0 to 5 within essential oils. This scale was chosen in accordance with the common range of aroma wheels used by professional brewers, and panelists were trained accordingly [18]. For these scents, the following products were used as control olfactory markers for the training process: respectively, standard mixes of (E)-β-caryophyllene and zingiberene (spicy), α-selinene and α-humulene (woody), β-selinene (herbal), (E)-β-farnesene and myrcene (citrus), and 3-mercaptohexanol (tropical fruit). These standards mixes were diluted at 5 different concentrations (20 µL·mL −1 , 50 µL·mL −1 , 100 µL·mL −1 , 150 µL·mL −1 , and 200 µL·mL −1 ) to ensure panelists exposure to different levels of aroma intensities. The higher concentrations used in this study are commonly considered to lead to strong olfactive detection for myrcene [19]. In the case of a scent expressed by two products (e.g., woody), both were diluted at equal concentrations to create the desired aroma. The range described also included a blank corresponding to theoretical zero on the defined aroma scale. In these samples on which the panelists were trained for twenty-five hours spread out weekly over six months, it was asked that the intensity of the corresponding scent be determined from 0 to 5. The different mixes at different concentrations were randomly distributed upon panelists at the beginning of each session. The panelists' efficiency regarding repeatability was deemed satisfactory for a standard deviation of less than 1.5. Panelists unable to provide such efficiency were eliminated; the other twenty became "selected assessors". Other exercises were conducted where panelists were given all 6 tubes from a randomly attributed scent and asked to sort them from lower to higher intensity.
At the end of the training, the twenty remaining panelists were asked to determine the intensity of the five scent categories on the thirty-nine essential oils. To avoid bias and olfactory saturation, these analyzes were triplicated and spread out over ten sessions. In these sessions, the 117 samples were randomly and blindly distributed.
Two-way ANOVAs were carried out between the ratio of the seven discriminant compounds described above and parameters such as year, proximity to the sea, and pedology. The corresponding p-values were considered significant when inferior to 0.05.
Additionally, the raw data corresponding to the abundance of the fifty-nine volatile compounds detected in the essential oils (C1-C59) for the thirty-nine Corsican accessions (1-39) were normalized using the centering normalization from the "scale" function in R. These data were then put in a dissimilarity matrix using Euclidean distance ("dist" function in R) and the hierarchical tree resulting from the complete matrix linkage ("hclust" function in R) was graphically plotted as a dendrogram.
The relative abundances of these compounds averaged in Table 3 are different from those exposed in a recent study on Italian wild hops [23], especially for the zingiberene, which is not detected in Italian population. These differences highlight a singular volatile profile of Corsican wild accessions among Mediterranean hops. Moreover, the very high standard deviations are evidence of strong dissimilarities between the samples, which means that there is a great quantitative variability within the studied territory.

Statistical Analyses-Year-to-Year Variability
Chemical characterization of essential oils shows a great quantitative variability. To determine the variables that can be described as discriminant factors influencing these dissimilarities, several batches of statistical analyses were performed using the RStudio software. In the PCA performed and illustrated in Figure 1, the thirty-nine accessions (1-39) are shown in black, and the fifty-nine volatile compounds as variables (C1-C59) are shown in blue. The two dimensions selected express a total of 72.5% of total variance, which seems fairly significant considering the diversity of the parameters and the size of the data set analyzed.
A differentiation emerges between two groups of accessions according to their abundances in (E)-β-caryophyllene (C9), β-selinene (C17), and α-selinene (C20) for the first group, and in myrcene (C2), (E)-β-farnesene (C12), and α-humulene (C13) for the second. Zingiberene's (C18) discrimination pattern is difficult to read because of the influence of other minor compounds that are not well represented by the two dimensions shown here. Therefore, another analysis was performed, focusing on the seven compounds mentioned above (Figure 2). In this figure, the thirty-nine accessions (1-39) are shown in black, and the seven volatile compounds as variables are shown in blue.
The PCA represented in Figure 2 seems to show a new differentiation pattern between four groups of accessions according to their abundance in β-selinene (C17) and α-selinene (C20) for the first group, (E)-β-farnesene (C12) and α-humulene (C13) for the second, myrcene (C2) for the third, and possibly zingiberene (C18) for the fourth. This last group is only composed of the accession Corte 2020 (2). These observations could depict a great chemical variability within Corsican wild hops.
This biplot also shows a great dissimilarity between some samples from the same locations but harvested over different years (e.g., Corte (1, 2, 3); Canale di Verde (4, 5, 6)) and could, therefore, highlight a temporal variability. However, it is important to mention that the significance of this PCA is not optimal considering the relatively poor representativity of the two dimensions (68.7%) expressed in this plot. Moreover, in this particular projection, the variables are relatively close to the center, indicating that their representation following

Statistical Analyses-Geographical Diversity
To comprehend geographical diversity, another principal components analysis was performed. Potential biases caused by year-to-year variability were limited by averaging the chemical profiles of the accessions depending on their harvesting site. The biplot obtained from the whole volatile profile is illustrated in Figure 3. In this figure, the thirteen locations are shown in black, and the fifty-nine volatile compounds as variables (C1-C59) are shown in blue.
locations but harvested over different years (e.g., Corte (1, 2, 3); Canale di Verde (4, 5, 6)) and could, therefore, highlight a temporal variability. However, it is important to mention that the significance of this PCA is not optimal considering the relatively poor representativity of the two dimensions (68.7%) expressed in this plot. Moreover, in this particular projection, the variables are relatively close to the center, indicating that their representation following these principal components is not great. Year-to-year chemical variability shows little to no clear pattern and suggests that essential oils compositions are certainly influenced by localized climatic phenomena.

Statistical Analyses-Geographical Diversity
To comprehend geographical diversity, another principal components analysis was performed. Potential biases caused by year-to-year variability were limited by averaging the chemical profiles of the accessions depending on their harvesting site. The biplot obtained from the whole volatile profile is illustrated in Figure 3. In this figure, the thirteen locations are shown in black, and the fifty-nine volatile compounds as variables (C1-C59) are shown in blue.  In this representation, Corte, Sortipiani, and Venzolasca appear to show great similarity following their abundance in β-selinene (C17), zingiberene (C18), and α-selinene (C20). These three accessions appear to grow at medium altitude in the same river basin (Tavignanu). Urbino, Lucciana, and Canale di Verde profiles seem highly comparable depending on their abundance in (E)-β-farnesene (C12). To a lesser extent, Casaperta and Santa-Maria-Poggio show high similarity following their abundance in myrcene (C2). The other accessions seem equally impacted by the seven discriminant compounds. These results allow us to theorize a possible variation in olfactive bouquets of essential oils obtained from the four groups described above.
As shown in Table 4, noteworthy high contents of zingiberene are only found in hops growing at medium altitude (130-400 m) away from the coast, which are areas characterized by specific pedoclimatic conditions (e.g., salinity).
The essential oils were submitted to a trained panel for olfactory and sensory analyzes. The averaged results are shown in Figure 4. Tropical fruit notes were studied by the panel as a common olfactory marker of hops but were not detected in Corsican wild hops, nor was 3-mercaptohexanol (a usual marker of this note) among the constituents of essential oils (Table A1). The blind olfactive analyzes of essential oils were triplicated, then all the scores from 0 to 5 for each accession were averaged and represented into aroma wheels.  To illustrate more clearly the similarities and dissimilarities between these stations and correlate them to environmental parameters, another statistical analysis was performed on the essential oils' chemical composition using a dissimilarity matrix (Euclidean distance) and depicted as a dendrogram ( Figure 5).
In Figure 5, two groups of samples (essential oils) share 78% of similarity. The first group is composed of the four accessions of Corte, Sortipiani, Casaperta, and Venzolasca, all of them located at medium altitude (80-400 m) inside the land. The second group is composed of the other accessions, all of them growing at low altitude levels (0-30 m) near the coastline. This observation could demonstrate the influence of altitude and possibly of salinity on the EOs profiles from Corsican wild hops. Within this group, a second differentiation emerges, with 54% of dissimilarity between Ajaccio, Porto-Pollo, and Propriano, all three located in the south-west part of the island on granitic soils, and the other accessions located in the north-east part of the island on sedimentary soils (Table  1). This observation could also highlight the influence of pedology on the EOs from Corsican wild hops. For readability purposes, stations were grouped by type of aromatic profile according to their scores for the five studied scents. These profiles corroborate the estimated fragrance notes from chemical composition and the literature, presented in Table 4. Here again, three groups are formed (Figure 4): spicy and herbal and woody notes; medium notes with an equilibrated ratio between all scents except "tropical fruit"; and herbal and citrus notes. These results confirm the relation between the chemical composition and the olfactive profiles of Corsican hops, which both appear to be influenced by pedoclimatic conditions, altitude, and distance to the seashore.
To illustrate more clearly the similarities and dissimilarities between these stations and correlate them to environmental parameters, another statistical analysis was performed on the essential oils' chemical composition using a dissimilarity matrix (Euclidean distance) and depicted as a dendrogram ( Figure 5).

Statistical Analysis-All Environmental Factors
To comprehend the compounds that are influenced by environmental factors, twoway ANOVAs were carried out between the seven discriminant compounds described earlier and parameters such as year, proximity to the sea, and pedology. The corresponding p-values are shown in Table 5. In Figure 5, two groups of samples (essential oils) share 78% of similarity. The first group is composed of the four accessions of Corte, Sortipiani, Casaperta, and Venzolasca, all of them located at medium altitude (80-400 m) inside the land. The second group is composed of the other accessions, all of them growing at low altitude levels (0-30 m) near the coastline. This observation could demonstrate the influence of altitude and possibly of salinity on the EOs profiles from Corsican wild hops. Within this group, a second differentiation emerges, with 54% of dissimilarity between Ajaccio, Porto-Pollo, and Propriano, all three located in the south-west part of the island on granitic soils, and the other accessions located in the north-east part of the island on sedimentary soils (Table 1). This observation could also highlight the influence of pedology on the EOs from Corsican wild hops.

Statistical Analysis-All Environmental Factors
To comprehend the compounds that are influenced by environmental factors, two-way ANOVAs were carried out between the seven discriminant compounds described earlier and parameters such as year, proximity to the sea, and pedology. The corresponding p-values are shown in Table 5. In Table 5, the influence of the year of growth is only considered significant (p-value < 0.05) on the (E)-β-farnesene abundance. This could mean that the annual variability, probably induced by seasonal climatic conditions, can influence its concentration in hops volatile fraction and, by extension, the expression of a herbal-citrus aroma profile. This could mean that, for a same cultivar, the herbal-citrus predominance in a Corsican hopped beer might vary depending on the year of harvest. It is important to note that hops follow a strict annual cycle, with a vegetative state during winter season. Therefore, cones produced every year might be greatly influenced by seasonal changes observed at such small temporal scale. Future work must focus on the study of correlation between annual climatic data (such as pluviometry, mean temperature, or drought extent) and chemical composition of Corsican wild hops essential oils. Sampling over a longer period could also be considered.
The influence of proximity to the sea is significant for all compounds (especially α -selinene) except (E)-β-farnesene and zingiberene (p-values = 0.408 and 0.106, respectively). This result is unexpected regarding the high amounts of zingiberene detected in hops growing in altitude and relatively far from the coast (e.g., Corte). Therefore, altitude and salinity might not be the factors influencing the chemical particularity of these hops. On the other hand, the pedology factor is highly significant for all compounds (including zingiberene) except α-humulene and myrcene. Corte is a location with quite unique soil (schistous/granitic) and the great amounts of zingiberene could actually be dependent on pedology.
It was previously proven that the aroma profiles of Corsican hops' essential oils are influenced by their chemical composition. Table 6 summarizes the scents that are believed to be influenced by the studied environmental factors. Table 6. Environmental factors influencing the aroma profile of Corsican hops.

Scents
Year Proximity to the Sea Pedology The ANOVAs referred to in Table 5, as well as the other statistical analyzes presented in previous sections of this work, tend to prove the impact of environmental factors on the chemical composition of Corsican wild hops and, by extension, on their aroma profile.

Valorization in Food Industry
The chemical and sensory analyzes of the essential oils of Corsican hops have been of great importance in determining their brewing potential. Each growing station and their associated pedoclimatic conditions seem to exert an influence on the aroma profile. The abundance of zingiberene is remarkable. This compound, as well as its particularly desirable spicy profile, is found to a lesser extent in most commercial cultivars. Thus, commercially brewed Corsican hops could lead to the emergence of a highly typical organoleptic bouquet singular to Corsican beers and induced by these high zingiberene concentrations. In addition, their harvest would allow breweries outside Corsica to obtain ginger-spicy beers.
The impact of terroir on the aroma of Corsican hops has been described in this study. However, the brewing process is known to thermally and microbiologically alter certain organoleptic compounds [24], and further studies are needed to confirm these possible impacts on the aroma of beers. In association with local breweries, future prospections could be conceptualized on beer samples made with the thirteen different accessions used in this article. A new panel could then be formed and trained for sensory analysis of beverages. It is important to specify that the correlation between the aromatic profile of Italian hops and that of the beers brewed with them is already known in the food industry [15].
The industrial development of the 20th century led to a steady decline in the use of natural raw materials, particularly in the world of chemicals, pharmaceuticals, and the food industry. As climate change and the availability of resources become more problematic, to limit their environmental impact, today's consumers tend to rethink their consumption modes patterns through local economy and the use of proximity products. The brewing world is no exception to this remodeling. By depicting their aroma profile, this article might enable Corsican breweries to use local wild hops as raw material. Following this work, we created the very first Corsican hop farm near the north coastline of the island. More than 2000 rhizomes were planted in the immediate vicinity of a brewery, which made it possible to study the impact of Corsican terroir on the cultivation of wild hops and foreign cultivars [11].

Conclusions
The aim of this study was to determine precisely and for the first time the variability of essential oils and the aroma of Corsican hops through both chemical and sensory analyzes. First, a sampling strategy on the scale of the island and the construction of the first hop farm in Corsica were initiated, with the intention of highlighting the typicality of Corsican hops and the impact of Corsican terroir on these different aspects.
Chemical and statistical analysis reveal a volatile profile that is quantitatively different from that of Italian hops, but also significant geographical and time-related diversity within Corsican hops where the main Eos' compounds are β-selinene, zingiberene, myrcene, and α-selinene. The particularity of Corsican wild hops' essential oils has been highlighted as well as their great interest due to the abundance of zingiberene, a compound with a ginger-spicy aroma highly valued in the brewing industry.
Beyond the substantial academic content, these advances allow the immediate economic and cultural enhancement of this local resource and could make Corsica the springboard for a Mediterranean brewing industry based on a guarantee of organoleptic typicality. Furthermore, this investigation demonstrates the importance of the hops cultivation's parameters for producing beers with the desired organoleptic notes.       Hexadecanoic acid 1961 tr-9.9 1.9 1.2 a : Literature non-polar retention indices [25]; b : experimental non-polar indices; c : experimental polar indices; italic: compounds showing significant variations.