Sensory profiling of honeybush tea (Cyclopia species) and the development of a honeybush sensory wheel

Sensory profiling of honeybush tea (Cyclopia species) and the development of a honeybush sensory wheel

Food Research International 66 (2014) 12–22 Contents lists available at ScienceDirect Food Research International journal homepage: www.elsevier.com...

1MB Sizes 10 Downloads 36 Views

Food Research International 66 (2014) 12–22

Contents lists available at ScienceDirect

Food Research International journal homepage: www.elsevier.com/locate/foodres

Sensory profiling of honeybush tea (Cyclopia species) and the development of a honeybush sensory wheel K.A. Theron a, M. Muller a, M. van der Rijst b, J.C. Cronje c, M. le Roux c, E. Joubert a,b,⁎ a b c

Department of Food Science, Stellenbosch University, Private Bag X1, Matieland, Stellenbosch 7602, South Africa ARC Infruitec–Nietvoorbij, Private Bag X5026, Stellenbosch 7599, South Africa Laboratory for Ecological Chemistry, Department of Chemistry and Polymer Science, Stellenbosch University, Private Bag X1, Matieland, Stellenbosch 7602, South Africa

a r t i c l e

i n f o

Article history: Received 24 June 2014 Accepted 24 August 2014 Available online 30 August 2014 Keywords: Herbal tea Descriptive sensory analysis Sensory wheel Gas chromatography–olfactometry Aroma-active compounds Eugenol

a b s t r a c t Several Cyclopia species (Cyclopia sessiliflora, Cyclopia longifolia, Cyclopia genistoides, Cyclopia intermedia, Cyclopia subternata and Cyclopia maculata), used as honeybush herbal tea, were analyzed using descriptive sensory analysis in order to develop a generic honeybush sensory wheel. It was found that the “characteristic” sensory profile of honeybush could be described as “floral”, “sweet-associated”, “fruity”, “plant-like” and “woody” with a sweet taste and a slightly astringent mouthfeel, whereas other attributes defined differences in the sensory characteristics between the Cyclopia species. The species could be divided into three distinct groups: group A (C. sessiliflora, C. intermedia and C. genistoides) associated with “fynbos-floral”, “fynbos-sweet” and “plant-like” attributes, group B (C. longifolia and C. subternata) with “rose geranium” and “fruity-sweet” and group C (C. maculata) with “woody”, “boiled syrup” and “cassia/cinnamon”. The large sample set enabled the development of a generic honeybush sensory wheel, comprising of 30 attributes, organized into positive and negative attributes in primary and secondary tiers. Gas chromatographic-olfactometry of the aroma fraction of a sample of C. maculata with a prominent spicy aroma indicated a high level of eugenol, the only aroma-active compound that associated with a spicy aroma. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction Honeybush is a traditional South African herbal tea, produced from the leaves and stems of a number of Cyclopia species. Concerted effort since the 1990s to expand the honeybush industry included the development of international markets. The major importers are the Netherlands, Germany, United Kingdom and USA (Joubert, Joubert, Bester, De Beer, & De Lange, 2011). The primary commercial product is the “fermented” herbal tea, produced through a high temperature oxidation process. Processing conditions and equipment vary from processor to processor, resulting in a large variation of sensory quality and thus a loss of consumer confidence. Additionally, several Cyclopia species are used for herbal tea production, with the retail product usually consisting of a blend of Cyclopia species. The composition of the blend depends largely on production yield and availability and is not governed by the need to produce a consistent sensory profile. Blending different Cyclopia species, without taking into account their different sensory profiles, results in variable profiles and therefore also a likely loss of consumer confidence. Current production consists mainly of Cyclopia intermedia, Cyclopia genistoides and Cyclopia subternata. However, with demand exceeding supply an interest ⁎ Corresponding author at: ARC Infruitec–Nietvoorbij, Private Bag X5026, Stellenbosch 7599, South Africa. Tel.: +27 21 809 3444; fax: +27 21 8083430. E-mail address: [email protected] (E. Joubert).

http://dx.doi.org/10.1016/j.foodres.2014.08.032 0963-9969/© 2014 Elsevier Ltd. All rights reserved.

in other species (Cyclopia sessiliflora, Cyclopia longifolia and Cyclopia maculata) with commercial potential has developed, expanding the range of honeybush sensory profiles. While correct blending could improve product consistency, it could lead to a loss of the unique flavor associated with individual species. Unique sensory profiles could potentially be used to establish niche markets. Previously, descriptive terms used to describe the flavor of honeybush tea were rather broad-based and included sensory descriptors such as “sweet” and “honey-like” (Du Toit & Joubert, 1998, 1999). Cronje (2010) also used broad-based sensory descriptors such as “characteristic honeybush” and “rose geranium-like” in an attempt to differentiate the major flavor differences between species. This lack of specific sensory terminology to describe the unique flavor profile of each Cyclopia species contributes to the present challenge of effective quality control and the development of products with specific flavor profiles for niche markets. Following the recent development of a sensory wheel for another South African herbal tea, rooibos (Koch, Muller, Joubert, Van der Rijst, & Næs, 2012), the honeybush industry indicated the need for a similar quality control tool. Current regulatory control of the quality of honeybush makes no provision for aroma or flavor, except to stipulate that it should be “characteristic honeybush” without any clarification of this description (Anonymous, 2000). Therefore this study was conducted on six Cyclopia species to develop a “generic” honeybush

K.A. Theron et al. / Food Research International 66 (2014) 12–22

sensory wheel, comprising the major positive and negative flavor, taste and mouthfeel attributes of this herbal tea. In addition, gas chromatographic-olfactometry (GC–O) of C. maculata with a very prominent spicy aroma was conducted to determine the major aroma-active compounds of this species, in particular to identify compound(s) contributing to a spicy note. 2. Materials and methods 2.1. Honeybush samples A total of 58 honeybush samples, representing six Cyclopia species (C. sessiliflora, C. longifolia, C. genistoides, C. intermedia, C. subternata and C. maculata), were randomly sourced from commercial processors. In cases where limited or no production samples of a species were available, additional samples were prepared by laboratory-scale processing at the Post-Harvest & Wine Technology Division of ARC Infruitec–Nietvoorbij, Stellenbosch, South Africa. Plant material was randomly harvested to introduce normal compositional variation. The full sample set thus represented a range of sensory qualities, typical of commercial honeybush. The samples of C. sessiliflora and C. longifolia consisted of seven independent batches each, whereas those of the other species consisted of eleven independent batches each, with each batch representing a replicate. For laboratory-scale processing different batches of plant material, comprising the shoots, were harvested at several locations in the Western Cape Province, South Africa during 2010 and 2011 and processed according to a standardized protocol as described by Le Roux, Cronje, Joubert, and Burger (2008). Briefly, the plant material, mechanically cut into small pieces, were moistened to ca. 65% moisture content and “fermented” at either 80 °C for 24 h or 90 °C for 16 h in temperaturecontrolled laboratory ovens. Following drying under controlled conditions (40 °C for 6 h) a sub-sample (200 g; b 10% moisture content) was sieved for 30 s using a SMC Mini-sifter (JM Quality Services, Cape Town, South Africa) and the fraction b12 mesh and N 40 mesh collected. The sieved plant material was stored in sealed glass jars at room temperature until analyzed. 2.2. Sample preparation Freshly boiled distilled water (900 g) was poured onto 11.25 g sieved plant material, infused for 5 min and strained into a 1 L stainless steel thermos flask. The infusion (ca. 100 mL) was served in white porcelain mugs covered with plastic lids to prevent loss of volatiles. Measures taken to keep the temperature of the infusions constant during serving include pre-heating of the thermos flasks and mugs and the use of a temperature-controlled (65 °C) scientific water baths during serving as proposed by Koch et al. (2012).

which all other samples could be compared, thereby allowing panelists to calibrate their sensory perception at the start of each training and testing session. A total of 68 aroma and 51 flavor, taste and mouthfeel descriptors were generated during the training period. By deliberating the relevance and redundancies of the descriptors, the list was reduced to 28 aroma, 23 flavor and 3 taste descriptors and 1 mouthfeel descriptor, totaling 55 descriptors (Table 1). The list of descriptors included positive, as well as negative sensory attributes, i.e. attributes associated with good and poor quality, respectively. A score sheet was then developed which was used by the panel to scale the intensity of each of the descriptors on a 100 mm unstructured line scale. After completion of training the assessors scored the intensities of the selected attributes for all samples. Each sample was analyzed only once. Two sessions were conducted per day during which 8 to 12 samples were analyzed. Samples were labeled with blinding codes and presented in a randomized order. The fixed point control sample was labeled as such so that it could be identified by the assessors and used for comparison. All analyses were conducted in booths fitted with controlled lighting and the data captured, using Compusense five® (Compusense, Guelph, Canada).

2.4. Gas chromatography–olfactometry Sample preparation and gas chromatography–olfactometry (GC–O) were carried out largely as described by Le Roux, Cronje, Burger, and Joubert (2012). An infusion of one sample of C. maculata (Mac 3), chosen for its prominent spicy aroma, was prepared by adding boiling distilled water (130 mL) to 20 g of the plant material in an insulated flask. The flask was sealed immediately and the plant material was infused for 15 min while swirling the contents of the flask periodically, followed by filtering. For each analysis, 50 mL of the filtrate was transferred to a

Table 1 Final aroma, flavor, taste and mouthfeel attributes used for descriptive analysis. Aroma attributes

Descriptors

Flavor, taste and mouthfeel attributes

Descriptors

Floral

Fynbos-floral Rose geranium Rose/perfume Lemon Orange Cooked apple Apricot jam Cherry Plant-like Woody Rooibos Pine Walnut Coconut Cassia/ cinnamon Dusty Yeasty Medicinal Burnt caramel Rotting plant water Hay/dried grass Green grass Cooked vegetables Fruity-sweet Boiled syrup Caramel Honey Fynbos-sweet

Floral

Fynbos-floral Rose geranium Rose/perfume Lemon Orange Cooked apple Apricot jam Cherry Plant-like Woody Rooibos Pine Walnut Coconut Cassia/ cinnamon Dusty Yeasty Medicinal Burnt caramel Rotting plant water Hay/dried grass Green grass Cooked vegetables Sweet Sour Bitter

Fruity

Plant-like

2.3. Descriptive sensory analysis Nine experienced assessors were selected to participate in this study. The sensory panel was trained according to the consensus method as described by Lawless and Heymann (2010). The basic protocol of Koch et al. (2012) was used to analyze all the samples during 24 training sessions where the panel generated aroma, flavor, taste and mouthfeel terminology. Four to six samples were analyzed per session. Aroma was defined as the aromatics perceived through orthonasal analysis, flavor as the retronasal perception and taste as the basic taste modalities. Mouthfeel was described as the tactile sensation that occurred in the oral cavity (Ross, 2009). As for Koch et al. (2012), reference standards illustrating the respective major sensory attributes of the herbal infusions were introduced to the panel to clarify the meaning of each descriptor. For each session the panel also received one specific honeybush sample (C. intermedia), not part of the test samples, which served as a control sample throughout all sessions. This sample was selected to serve as a fixed point to

13

Nutty Spicy Negative

Sweetassociated

Fruity

Plant-like

Nutty Spicy Negative

Taste

Mouthfeel

Astringent

14

K.A. Theron et al. / Food Research International 66 (2014) 12–22

(a)

Positive aroma attributes

AFynbos-sweet

100

AFynbos-floral AWoody

90

APlant-like AFruity-sweet

80

AApricot jam

Percentage of samples

70

ARose geranium ABoiled syrup

60

ACassia/Cinnamon ARose/Perfume

50

ACaramel

40

ALemon AWalnut

30

ACoconut ACooked apple

20

AHoney

10

ARooibos APine

0

0

5

10

15

ACherry essence

20

AOrange

Average intensity (out of 100)

(b)

Negative aroma attributes

100

Percentage of samples

90 80

AHay/Dried grass

70

ABurnt caramel

60

ADusty

50

ARotting plant water

40

AGreen grass

30

AMedicinal

20

AYeasty

10

ACooked vegetables

0

0

1

2

3

4

Average intensity (out of 100)

(c)

Taste and mouthfeel attributes

100 90

Percentage of samples

80 70 60

TSweet Astringent TBitter TSour

50 40 30 20 10 0

0

5

10

15

20

25

30

Average intensity (out of 100) Fig. 1. Scatter plots showing the percentage of honeybush samples exhibiting a certain attribute vs. the average intensity of the specific attribute. The letters “A” and “T” in front of an attribute refer to aroma and taste attributes, respectively.

K.A. Theron et al. / Food Research International 66 (2014) 12–22

100 mL glass bottle with adapted cap (Burger, Marx, le Roux, & Burger, 2006), sealed, and incubated at 50 °C for 30 min, after which the volatile organic compounds (VOCs) in the headspace were enriched at 50 °C for 17 h using a SEP60 sample enrichment probe (MasChrom Analisetegniek, Stellenbosch, South Africa). The probe contained 60 mm polydimethylsiloxane (PDMS) tubing, equivalent to 56 mg of PDMS (Burger et al., 2006). GC–O analyses were performed on a conventional Carlo Erba HR gas chromatograph (Milan, Italy), converted for GC–O use by installing a glass effluent splitter, a humidified air conduit, and a glass sniffing port. Helium was used as the carrier gas at a linear velocity of 28.6 cm/s. The VOCs sorbed in the PDMS of the sample enrichment probe were desorbed at an injector temperature of 230 °C and analyzed on a capillary column (glass, 40 m × 0.25 mm i.d.) coated with 0.25 μm of PS-089-OH (DB-5 equivalent), using a temperature program of 2 °C/ min from 40 °C to 280 °C. The VOCs were subjected to GC–O evaluation by eight expert assessors who were required to individually sniff the GC effluent and report the results according to the detection frequency (DF) method. In order to prevent sensory “fatigue”, each assessor was required to sniff the effluent during alternating first and second halves of consecutive analyses. The total number of panel members who could positively detect an odorant at a specific retention time was expressed as a percentage of the total number of assessors. A compound was considered to be aroma-active if it was positively detected by at least 50% of the assessors (Áslaug & Rouseff, 2003).

15

2.5. Gas chromatography–mass spectrometry GC–MS was performed on a Carlo Erba QMD 1000 GC–MS system using the GC analytical parameters described by Le Roux et al. (2012). The line-of-sight interface and ion-source were kept at 250 and 180 °C, respectively. Electron-impact (EI) mass spectra were recorded at 70 eV at a scan rate of 0.9 s/scan, an interscan time of 0.1 s, and a scan range of 40–300 amu. GC–MS data processing was carried out using an NBS database (VG Masslab, VG Instruments, Manchester, UK) and NIST mass spectral library (version 2.0d, National Institute of Standards and Technology, Gaithersburg, MD, USA). The identities of the constituents under investigation were confirmed by retention time comparison with authentic commercially available reference samples. 2.6. Data analysis A complete block design was used for the sensory analysis. The six Cyclopia species each represented a treatment whereas each sample of a species was considered an independent replication. Panel performance was monitored using PanelCheck Software (www.panelcheck. com), while panel reliability was tested by subjecting the data to test– retest analysis of variance (ANOVA) using SAS® software (Version 9.2; SAS Institute Inc., Cary, USA). Judge ∗ Replication interaction and Judge ∗ Sample interaction were used as measures of the panel precision and homogeneity, respectively. The Shapiro–Wilk test was used

Biplot (axes F1 and F2: 28.03 %) 6

Mac3 ABoiled syrup FCassia/Cinnamon Mac2 FWalnut ACassia/Cinnamon Mac1 Mac8 Mac4 AWalnut

Mac5

ACooked apple

4 Sub10 TSweet Mac7 Mac6

ADusty ACoconut

FRose/Perfume Int10

F2 (11.97 %)

2

ARose/Perfume

Sub9

Int7

Mac9

ACherry essence Int8 Sub7 Mac10 Sub11 Lon4 FYeasty Mac11 Lon3 FMedicinal APine Sub1 FWoody AYeasty FDusty Int5 ARose geranium FCherry essence FRooibos Ses7 FRose geranium AMedicinal Int4 Sub5 AHoney Int9 Sub4 Lon2 Lon1 Int6 AFruity-sweet FLemonLon7 Int11 FHay/DriedgrassSub3 FRottting plantwater Sub8 Gen3 FApricot jam Int2 ARooibos Int1 FCooked vegetables AApricot jam Int3 Gen6Gen1 ACooked vegetables ACaramel Sub2 Gen2 Gen11 ALemon Sub6 Lon6 ARotting plantwater AHay/Dried grass FGreen grass FBurnt caramel AFynbos-sweet Gen5 ABurnt caramel Gen8 Ses3 Ses6 AFynbos-floral Ses1 Gen4 Ses4 FFynbos-floral FPlant-like AGreen grass TBitter Gen7

0

-2

-6

-4

-2

0

Gen10

TSour Astringent

APlant-like

-8

Lon5

Gen9

Ses2

Ses5

-4

-6 -10

AWoody

2

4

6

8

10

F1 (16.06 %) Fig. 2. PCA bi-plot showing the positioning of the 58 honeybush samples relative to both positive and negative sensory attributes. The letters “A”, “F” and “T” in front of the attributes refer to aroma, flavor and taste attributes, respectively. The abbreviations Ses, Lon, Gen, Int, Sub and Mac refer to the specific Cyclopia species; C. sessiliflora, C. longifolia, C. genistoides, C. intermedia, C. subternata and C. maculata, respectively.

16

K.A. Theron et al. / Food Research International 66 (2014) 12–22

to test for non-normality of the residuals (Shapiro & Wilk, 1965). In the event of significant non-normality (p ≤ 0.05) outliers were identified and residuals larger than 3 were removed. Principal component analysis (PCA) using the correlation matrix was conducted using XLStat (Version 7.5.2, Addinsoft, New York, USA). Discriminant analysis (DA) was used to differentiate between the Cyclopia species.

taste and a slight astringent mouthfeel. The contribution of minor notes to the overall sensory profile should, however, not be totally disregarded (Ryan, Prenzler, Saliba, & Scollary, 2008).

3. Results and discussion

The PCA bi-plot (Fig. 2) displays the association between the sensory attributes, as well as the positioning of the samples relative to each other. According to the first principal component (PC1), most of the negative attributes are located in the two right quadrants of the PCA plot, whereas most of the positive attributes are located on the left (Fig. 2). According to PC2, the samples in the bottom left quadrant are associated with sensory attributes such as “fruity-sweet”, “apricot jam”, “fynbos-floral”, “fynbos-sweet”, “plant-like”, “green grass” and “hay/dried grass”, whereas the sensory attributes “cassia/cinnamon”, “woody”, “coconut” and “boiled syrup”, all located in the top right quadrant, associate with C. maculata samples (Fig. 2). Fig. 2 furthermore indicates that there was some degree of grouping or clustering of the samples based on the Cyclopia species. In order to generate a perceptual map of potential groupings, as well as determining which of the sensory attributes are responsible for these groupings, DA was carried out using only the positive sensory attributes. The resulting DA plot (Fig. 3) shows that three groupings could effectively be identified: Group A (C. sessiliflora, C. genistoides and C. intermedia), Group B (C. longifolia and C. subternata) and Group C (C. maculata). In order to determine which sensory attributes caused these groupings, separate PCA plots were generated of the positive aroma, taste and mouthfeel attributes (Fig. 4), as well as the negative aroma attributes (Fig. 5). Bar graphs indicating the differences between the Cyclopia species in terms of these attributes are depicted in Fig. 6. Only attributes with mean intensity scores N5 were included in Fig. 6. “Burnt caramel”, “green grass” and “hay/dried grass” aromas were therefore included because the average values for specific Cyclopia species were N5 (Fig. 6b), indicating that these attributes may be important in describing the sensory profiles of those species. For the purpose of comparison only the aroma attributes

During training a total of 119 sensory descriptors were generated. For the purpose of efficient sensory profiling it is necessary to reduce the number of descriptors to about 10 to 20 (Vannier, Brun, & Feinberg, 1999). However, since the present study was the first attempt to profile the sensory attributes for honeybush it was important to capture the full sensory variation of the different honeybush species. Rigid reduction of descriptors to meet the limits proposed by Vannier et al. (1999), however, could result in a loss of specific attributes that would be essential in characterizing the unique sensory profiles of the respective Cyclopia species. The drawback of working with a strongly reduced set of descriptors has been noted by other authors (Stone & Sidel, 1985; Wolters & Allchurch, 1994). The comprehensive list of original descriptors compiled for honeybush was thus reduced to 28 aroma, 23 flavor and 3 taste attributes, as well as one mouthfeel attribute. The relative importance of the major sensory attributes in honeybush is visualized in the graphs depicting the percentage of samples that exhibit a specific attribute plotted against its average intensity (Fig. 1). These graphs give an indication of the perceived intensity of the sensory attributes in an infusion, as well as their prevalence amongst the samples. The aroma attributes (Fig. 1a) “fynbos-sweet”, “fynbos-floral” and “woody” obtained the highest scores, followed by “plant-like”, “fruitysweet” and “apricot jam”. These six positive aroma attributes were perceived in at least 80% of the honeybush samples. “Rose geranium”, “boiled syrup” and “cassia/cinnamon” aroma occurred in more than 60% of the samples. Other positive attributes were not perceived as frequently and at much lower intensities. In terms of the negative aroma attributes (Fig. 1b), only “hay/dried grass” was perceived in more than 50% of the samples at extremely low intensities. Sweet taste and astringency were perceived in all the honeybush samples (Fig. 1c) at low and extremely low intensities, respectively. Bitter and sour tastes were detected in more than 90% of the honeybush samples, however, also at extremely low intensities (b 5). Even though some of the sensory attributes occurred in less than 80% of the samples, and at low average intensities, it does not necessarily mean that they should be disregarded. These attributes may be those which are more species-specific. Also, attributes rated low in intensity can still contribute significantly to the aroma or flavor of the infusion, especially the negative sensory attributes. Since only the average intensity values are reflected in Fig. 1, it is also important to examine the range of intensity values in order to gain a better understanding of the importance of certain attributes. Some attributes had high maximum intensity scores, but were only present in a small number of samples. The attributes, “cassia/cinnamon” and “hay/dried grass”, for instance, had maximum aroma intensities of 53 and 34, respectively. However, while present in more than 60% of the samples only three out of 58 samples had intensities higher than 33 for “cassia/cinnamon”. For “hay/dried grass” aroma only one sample had an intensity of more than 16. For the attributes “lemon” and “coconut”, only one sample had intensity scores of 11 and 16, respectively, whereas all other samples had negligible intensity scores for these attributes (b5 on a 100-point scale). The latter attributes can therefore not be considered as “characteristic” of all batches of honeybush since they were only detected in a very small percentage of samples. Based on intensity and occurrence, the overall sensory profile of honeybush can thus be described as having prominent “sweet-associated”, “floral”, “fruity”, “plant-like” and “woody” aroma and flavor notes with a sweet

Observations (axes F1 and F2: 69.84 %) 6

GROUP A 4

Ses Gen Int

2

F2 (22.20 %)

3.1. Sensory attributes

3.2. Segmentation of sensory profiles of Cyclopia species

0

-2

Lon Mac

Sub

-4

GROUP C GROUP B -6 -10

-8

-6

-4

-2

0

2

4

6

8

F1 (47.64 %) Fig. 3. DA plot illustrating honeybush sample groupings based on sensory attributes. The abbreviations Ses, Lon, Gen, Int, Sub and Mac refer to C. sessiliflora, C. longifolia, C. genistoides, C. intermedia, C. subternata and C. maculata, respectively.

K.A. Theron et al. / Food Research International 66 (2014) 12–22

and taste and mouthfeel attributes were examined since the aroma and flavor attributes were highly correlated for most attributes. 3.2.1. Group A The first group consists of C. sessiliflora, C. genistoides and C. intermedia (Fig. 3). Most of these samples are positioned in the upper quadrants of the PCA bi-plot, associating with the attributes “fynbos-sweet”, “fynbos-floral”, “lemon”, “plant-like”, sour, bitter and astringent (Fig. 4). A few samples situated in the top right quadrant are associated with “caramel” and “woody”. Although these species had very similar sensory characteristics, intensity differences are noticeable in Fig. 6. C. sessiliflora had the highest average intensities of “plant-like” and “green grass” aromas. C. genistoides had the highest average intensity for “apricot jam” aroma. For most aroma attributes C. intermedia did not differ significantly from the other two species within this group. As for taste and mouthfeel attributes, C. intermedia had a significantly higher average score for sweet taste than C. genistoides, while C. sessiliflora and C. genistoides had the highest intensities for sour and bitter, respectively (Fig. 6c). C. genistoides was significantly more astringent than C. intermedia. In summary, C. sessiliflora samples were associated with “green” aroma (“plant-like” and “green grass”), as well as a sour taste, and C. genistoides with “apricot jam” aroma and a slightly bitter taste. C. intermedia did not display a distinct profile as most positive attributes

17

were perceived at moderate intensities. C. genistoides was often related to the negative attributes, “burnt caramel” and “hay/dried grass”, and C. sessiliflora to “green grass”. It is likely that the specific negative attributes of these samples can be attributed to poor processing. “Burnt caramel” and “green grass” may be a result of over- and under-fermentation, respectively. Du Toit and Joubert (1999), investigating fermentation conditions, demonstrated that over-fermented C. intermedia associated with “burnt caramel” notes. 3.2.2. Group B Group B is composed of C. longifolia and C. subternata (Fig. 3) and again it can be assumed that these species have reasonably similar sensory profiles. Most of the samples are located in the bottom left quadrant of the PCA bi-plot (Fig. 4), and both species are therefore associated with attributes such as “rooibos”, “apricot jam”, “rose geranium”, “fruity-sweet”, “rose/perfume” and sweet taste. There are a few exceptions: Two C. longifolia samples in the top right quadrant correlated with “woody” and “caramel”, whereas two C. subternata samples situated in the bottom right quadrant associated with “walnut” and “coconut”. Although C. longifolia and C. subternata have similar sensory characteristics, there are also differences (Fig. 6). These species had high average intensities for “fruity-sweet” and “rose geranium”. C. longifolia had a low mean intensity for “fynbos-floral”. Some of the C. longifolia samples had an unpleasant “rotting plant water” aroma which may have

Biplot (axes F1 and F2: 32.93 %) 6

Astringent TSour Ses6 APlant-like

4

Ses2

FPlant-like ACaramel Gen11 Sub11 Int5

Ses5 TBitter Ses4 Gen4 Gen7 Ses1 Ses3 ALemon Gen1 FFynbos-floral Gen5 Gen6Int1 AFynbos-floral Lon7 Int11 Gen2 Sub6 Int3 Int9 AFynbos-sweet

F2 (15.11 %)

2

0

Gen10

Gen9

Lon6 Int6 Ses7

Lon5

FWoody

AWoody Mac10

APine FCooked apple Gen8 Mac11 Mac9 ARooibos Lon1 Int4 Sub2 Sub8 Int8 Gen3 Int2 FRooibos FLemon ACherry essence Lon2 Mac4 FApricot jam Sub4 AHoneyInt10 Mac1 ACoconut Int7 Sub3Sub5 Mac2 Sub9 AApricot jam Sub1 Lon3 Sub10 ACassia/Cinnamon Sub7 FRose geranium ABoiled syrup Mac8 Mac6 AWalnut Lon4 AFruity-sweet ARose geranium FCassia/Cinnamon Mac7 FRose/Perfume FWalnut ARose/Perfume Mac3 ACooked apple

-2

-4

TSweet Mac5 -6

-7

-5

-3

-1

1

3

5

7

F1 (17.82 %) Fig. 4. PCA bi-plot showing the positioning of the 58 honeybush samples relative to the positive sensory attributes including all taste attributes. The abbreviations Ses, Lon, Gen, Int, Sub and Mac refer to the specific Cyclopia species; C. sessiliflora, C. longifolia, C. genistoides, C. intermedia, C. subternata and C. maculata, respectively. The letters “A”, “F” and “T” in front of the attributes refer to aroma, flavor and taste attributes, respectively.

18

K.A. Theron et al. / Food Research International 66 (2014) 12–22

resulted from poor processing. In terms of sweet, bitter, sour and astringency the two species had similar intensities (Fig. 6c). Overall it is clear that there are not prominent sensory differences between C. longifolia and C. subternata.

not previously been used to describe honeybush. In order to determine which compound, or combination of compounds, is responsible for this unique aroma note, a prominent “spicy” C. maculata sample (Mac3) (Fig. 7) was selected and analyzed, using GC–MS and GC–O. A list of the aroma-active compounds found in the sample and the aroma attributes reported to be associated with these compounds are summarized in Table 2. A number of aroma-active volatile compounds associated with “floral”, “fruity”, “woody” and “sweet” aromas were identified using GC–O. The only volatile component identified in Mac3 that had a spicy character was eugenol. The GC–O panel members described the peak as having a sweet, spicy or clove aroma. The prominent spice note of C. maculata could therefore possibly be attributed to the high relative concentration of eugenol (Table 2). Eugenol is described as having a “spicy, clove-like” aroma (Acree & Arn, 2004). Eugenol has previously been identified as an aroma-active compound in C. subternata (Le Roux et al., 2012). A comparison of the chromatograms of their respective volatile fractions showed that eugenol was a major peak in Mac3, while in C. subternata it was only a very minor peak (Supplementary information). The fact that the floral notes in C. maculata received low scores may be due to the overpowering and/or masking effect of the spicy note as perceived by the sensory panel (Fig. 7). Although certain compounds can be linked to specific aroma notes, it is important to realize that aroma notes of different compounds can mask or suppress one another, and combinations of aroma compounds

3.2.3. Group C Group C (Fig. 3) comprises only C. maculata, indicating that this species had very different sensory attributes. C. maculata is located in the bottom right quadrant of the PCA bi-plot (Fig. 4). This reflects a strong association with attributes such as “cassia/cinnamon”, “boiled syrup”, “walnut”, “cooked apple” and “coconut”. C. maculata had low average scores for “fynbos-sweet”, “fynbos-floral”, “rose geranium” and “plant-like” aromas (Fig. 6a). It scored highest for “boiled syrup” and “woody” aroma, although in the latter case it only differed significantly (p ≤ 0.05) from C. longifolia. The most prominent difference between C. maculata and the other species was the “cassia/ cinnamon” aroma noted in some of the C. maculata samples (Fig. 6b). Except for “hay/dried grass” aroma there were no other prominent negative attributes associated with the C. maculata samples. In terms of the taste attributes, C. maculata scored similarly than the other Cyclopia species for sweet taste and low for sour, bitter and astringency (Fig. 6c). C. maculata could thus be distinguished from the other species mainly because of its relatively strong “cassia/cinnamon” aroma. The presence of this spicy note was unexpected as this sensory attribute has

Biplot (axes F1 and F2: 40.10 %) 8

Int10 AYeasty Int9 FYeasty

6

FDusty AMedicinal FMedicinal

F2 (15.07 %)

4

Int5

Int7 ADusty

2

Int8 Mac7 Mac2 Sub7 Mac5 Sub11 Mac1 Int11 Gen6 Sub4Lon1 Gen1 Sub1 Lon4 Gen5 Int4 Sub6 ASour Mac11 Sub10Sub8 Sub9 Int2 Lon7 Ses5 Ses6 Mac8 Sub2 Int1Sub3Mac4 Mac9 Sub5Lon2 Int6 Ses7Gen3 Lon3 Mac6 Lon6 Int3 Gen4 Mac3 Mac10 Ses4 Gen8 AHay/Dried grass Ses2 Ses1 ACooked vegetables Ses3 Gen7 FHay/Driedgrass Gen2

0

-2

Gen9

FBurnt caramel ABurnt caramel Gen10 FCooked vegetables ARotting plantwater FRottting plantwater Gen11

Lon5

AGreen grass FGreen grass

-4

-8

-6

-4

-2

0

2

4

6

8

F1 (25.03 %) Fig. 5. PCA bi-plot showing the positioning of the 58 honeybush samples relative to the negative sensory attributes. The abbreviations Ses, Lon, Gen, Int, Sub and Mac refer to the specific Cyclopia species; C. sessiliflora, C. longifolia, C. genistoides, C. intermedia, C. subternata and C. maculata, respectively. Rotting = Rotting plant water, Cooked veg = Cooked vegetables. The letters “A” and “F” in front of the attributes refer to aroma and flavor attributes, respectively.

K.A. Theron et al. / Food Research International 66 (2014) 12–22

can produce new aroma characteristics (Naudé & Rohwer, 2013). Also, certain compounds which are present in concentrations below their odor threshold or which have no odor activity when assessed individually can contribute to the aroma when they are in a mixture (Delahunty, Eyers, & Dufour, 2006). Furthermore, the role of non-aroma active compounds and/or sub- and peri-threshold odorants in the modification of the sensory perception of aroma-active compounds cannot be ignored as postulated by Ryan and co-workers (Ryan et al., 2008).

19

3.3. Sensory wheel In order to create a honeybush sensory wheel, the number of attributes was further reduced to 30 terms (26 flavor, 3 taste and 1 mouthfeel) based on frequency and intensities. These terms were assembled to form a three-tiered wheel (Fig. 8). Starting from the middle, forming the inner tier, are the specific attributes such as “fruity-sweet”, “boiled syrup” and “caramel”. The second tier comprises ten primary C. sessiliflora

30

C. genistoides

(a)

C. intermedia

25

C. subternata

a

a

ab

20 bc

a abc

ab

C. maculata

ab

abc

ab ab

ab

bc

c c

15

C. longifolia

a

abc cd

10

c

d

bc

a ab

a a

bc

bc

c

a

a

ab

bc

a

c

5

bc

a

0 AFynbos-sweet

AFynbos-floral

ARose geranium

ARose/Perfume

AFruity-sweet

30

C. sessiliflora C. genistoides C. intermedia C. subternata C. longifolia C. maculata

(b) 25

a a

20

15

a

a ab ab

10

ab

AApricot jam

b ab

a

bc

bc

b

ab c

c

5

b

b b b

b

b

ab

ab

b

b

a

a

b b b

b

b

b

ab ab

ab b

b b b b

0 AWoody

APlant-like

ACassia/Cinnamon ABoiled syrup

ABurnt caramel

30

(c) 25

a a

AGreen grass

AHay/Dried grass

C. sessiliflora C. genistoides C. intermedia C. subternata C. longifolia C. maculata

a a

ab b

20

15 a

10

ab a bc c c

a b

5

b

b b

b

c

b

bc c bc

c

0 TSweet

TBitter

TSour

Astringent

Fig. 6. Average attribute intensities for aroma (a, b), taste and mouthfeel attributes (c) of six Cyclopia species. Bars with different alphabetical letters are significantly different from each other (p ≤ 0.05). The letters “A” and “T” in front of the attribute name refer to aroma and taste, respectively. Only the 13 most important aroma attributes are shown based on their average values.

20

K.A. Theron et al. / Food Research International 66 (2014) 12–22

Aroma of C. maculata (Mac3)

Table 2 (continued)

AFynbos-floral ABurnt caramel

40

Compound namea

RIb

Unidentified (E)-β-Damascone

1400 1408 A

Unidentified 2,3-Dehydro-γ-ionone (E)-β-Ionone Unidentified Unidentified Bovolide

1454 1462 B 1485 A 1492 1515 1523 B

Unidentified C269 [m/z 220 (M+ C15H24O+, 2%)] 3-Hydroxy-α-damascone Unidentified Unidentified

1600

ARose geranium

30

ADusty

AApricot jam

IDc DFd (%)

20 10

ACoconut

ACooked apple 0

AWalnut

AWoody

ACassia/cinnamon

APine

ABoiled syrup

AFruity-sweet

1608 B 1658 1709

Aroma descriptor

earthy green-florale – Fruity (apple-citrus), tea-like with slight minty notese 75 – 100 – 100 Woody, fruityo 50 – 62.5 – 100 Celery- and lovage-like, fruity and pleasante 100 – 100 100

50 – 62.5 – 62.5 –

a

Fig. 7. Spider plot showing the mean scores for the aroma attributes associated with C. maculata (Mac3) as determined by descriptive analysis. The letter “A” in front of the sensory attribute refers to aroma.

Table 2 Aroma-active compounds in the volatile fraction of Cyclopia maculata (Mac3). Compound namea

RIb

IDc DFd (%)

Aroma descriptor

3-Methylbutanoic acid

880 A

100

Hexanal (R)-2-Methylbutanoic acid 1-Octen-3-ol 6-Methyl-5-hepten-2-one

783 890 971 979

A A A A

100 100 100 87.5

(E,E)-2,4-Heptadienal Phenyl acetaldehyde (E)-2-Octenal Terpinolene cis-Linalooloxide

1000 1033 1047 1073 1079

A A A A A

62.5 100 50 50 75

(E,E)-3,5-Octadien-2-one (R/S)-Linalool Nonanal 2-Phenylethanol

1080 1093 1100 1121

A A A A

100 75 100 87.5

(E,Z)-2,6-Nonadienal

1143 A

100

(E)-2-Nonenal

1150 A

100

α-Terpineol Safranal (E,E)-2,4-Nonadienal 2-Ethyl-3-methylmaleimide p-Anisaldehyde Geraniol

1186 1193 1213 1247 1253 1260

75 87.5 100 75 75 75

(E,E,Z)-2,4,6-Nonatrienal (E,Z)-2,4-Decadienal Geranyl formate Unidentified C162 [m/z 166 (M+ + C10H14O+ 2 , 100%), 95 (C7H11, 92%), 123 (C8H11O+, 68%)] Unidentified (E,E)-2,4-Decadienal

1267 B 1287 B 1293 A 1293

100 50 50 50

1307 1314 A 1327 B 1329 1343

62.5 – 87.5 Fried, waxy, fatty, orangelikee 100 – 50 – 100 –

1357 A 1371 B 1379 A

100 Warm-spicy, drye 62.5 Tobacco-likee 100 Woody, sweet, fruity,

Piperonal Unidentified Unidentified C178 [m/z 154 (M+ C9H14O+ 2 , 1%), 84 (C4H4O+ 2 , 100%)] Eugenol 2,3-Dehydro-α-ionone (E)-β-Damascenone

A A A B A A

Acid acrid, cheesy, unpleasante Green, grassy odorf Cheesy, sweaty, sharpe Mushroomg Oily-green, pungentherbaceous, grassy, with fresh and green-fruity notese Orange oil, oilyh Floral, sharpi Fatty, nuttyj Sweet-piney, oilye Earthy, floral, sweet, woodyj Fatty, fruity, mushroome Refreshing, floral-woodye Strong, fatty, orange, rosej Mild, warm, rose-honeylikee Green-vegetable, cucumber or violet leafe Green, cucumber, aldehyde, fattye Floral, sweet, lilac-typee Saffron-likek Fruity (citrus), fattyl – Sweet, floral, hay-likee Sweet, floral, rose, citruslikee Oat-likem Fattyn Fresh, green-rosy, fruitye –

In order of elution from apolar PS-089 column (DB-5 equivalent). Retention index, relative to C5–C18 n-alkanes, on PS-089 column (DB-5 equivalent). Identification: A, comparison of mass spectrum and RI with those of an authentic reference compound; B (tentative identification), HRGC–MS data and comparison of mass spectrum and RI with NBS and NIST databases and published data (Le Roux et al., 2012). d Detection frequency. e Aroma descriptors compiled from Cronje (2010). f Aroma descriptors compiled from Őműr-Őzbek and Dietrich (2008). g Aroma descriptors compiled from Acree and Arn (2004). h Aroma descriptors compiled from Jordán, Margaría, Shaw, and Goodner (2002). i Aroma descriptors compiled from Benn and Peppard (1996). j Aroma descriptors compiled from Chin, Eyres, and Marriott (2011). k Aroma descriptors compiled from Culleré, San-Juan, and Cacho (2011). l Aroma descriptors compiled from Piccino, Boulanger, Descroix, and Shum Cheong Sing (2014). m Aroma descriptors compiled from Schieberle and Schuh (2006). n Aroma descriptors compiled from Carunchia Whetstine, Croissant, and Drake (2005). o Aroma descriptors compiled from Pino (2014). b c

descriptors (“sweet-associated”, “nutty”, “spicy”, “herbaceous”, “fruity”, “floral”, “vegetative”, “chemical”, “earthy”, and taste and mouthfeel). These attributes are generic terms that group together similar, secondary descriptors found in the inner tier. The outermost tier of the sensory wheel groups the attributes as positive or negative attributes. Negative attributes were included as the wheel will serve as a quality control tool and their inclusion will assist honeybush processors to identify batches with poor quality. Concise definitions for the respective attributes are provided as supplementary information. The first generic sensory wheel developed for honeybush herbal tea in this study can be used as a communication tool between researchers, and industry role players. Additionally, it can be used for comparing and monitoring product quality and consistency, as well as for the development of new honeybush products. Future research will be necessary to fine-tune the wheel or to develop species-specific wheels.

4. Conclusions The “characteristic” sensory profile of honeybush can be described as a combination of “sweet-associated”, “floral”, “fruity”, “woody” and “plant-like” aromas with a sweet taste and a slightly astringent mouthfeel. The six Cyclopia species analyzed showed three distinct groups based on their sensory attributes, indicating that it would thus be possible to blend Cyclopia species of the same group without altering their typical flavor. It should, however, be kept in mind that even within a group there were subtle differences between species.

K.A. Theron et al. / Food Research International 66 (2014) 12–22

21

Dusty

Fig. 8. Generic sensory wheel for honeybush herbal tea, produced from Cyclopia species.

Acknowledgments This work is based on research supported in part by the National Research Foundation of South Africa (Grant UID 70525). The grant holder (EJ) acknowledges that opinions, findings and conclusions or recommendations expressed in any publication generated by the NRF supported research are those of the authors, and that the NRF accepts no liability whatsoever in this regard. The authors thank the honeybush industry for supplying plant material and the processed product. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.foodres.2014.08.032. References Acree, T., & Arn, H. (2004). Flavornet and human odor space. http://www.flavornet.org (Accessed March 2012) Anonymous (2000). Agricultural Product Standards Act. Act no. 119 of 1990. G.N.R. 1177/ 2000. Johannesburg, South Africa: Lex Patria Publishers. Áslaug, H., & Rouseff, R. L. (2003). Identification of aroma active compounds in orange essence oil using gas-chromatography–olfactometry and gas-chromatography–mass spectrometry. Journal of Chromatography A, 998, 201–211. Benn, S. M., & Peppard, T. L. (1996). Characterization of tequila flavor by instrumental and sensory analysis. Journal of Agricultural and Food Chemistry, 44, 557–566.

Burger, B. V., Marx, B., le Roux, M., & Burger, W. G. (2006). Simplified analysis of organic compounds in headspace and aqueous samples by high-capacity sample enrichment probe. Journal of Chromatography A, 1121, 259–267. Carunchia Whetstine, M. E., Croissant, A. E., & Drake, M. A. (2005). Characterization of dried whey protein concentrate and isolate flavour. Journal of Dairy Science, 88, 3826–3839. Chin, S., Eyres, G. T., & Marriott, P. J. (2011). Identification of potent odourants in wine and brewed coffee using gas chromatography–olfactometry and comprehensive twodimensional gas chromatography. Journal of Chromatography A, 1218, 7487–7498. Cronje, J. C. (2010). Chemical characterization of the aroma of honeybush (Cyclopia) species. (PhD Dissertation). South Africa: Stellenbosch University. Culleré, L., San-Juan, F., & Cacho, J. (2011). Characterisation of aroma active compounds of Spanish saffron by gas chromatography–olfactometry: quantitative evaluation of the most relevant aromatic compounds. Food Chemistry, 127, 1866–1871. Delahunty, C. M., Eyers, G., & Dufour, J. P. (2006). Gas-chromatography–olfactometry. Journal of Separation Science, 29, 2107–2125. Du Toit, J., & Joubert, E. (1998). The effect of pretreatment on the fermentation of honeybush tea (Cyclopia maculata). Journal of the Science of Food and Agriculture, 76, 537–545. Du Toit, J., & Joubert, E. (1999). Optimization of the fermentation parameters of honeybush tea (Cyclopia). Journal of Food Quality, 22, 241–256. Jordán, M. J., Margaría, C. A., Shaw, P. E., & Goodner, K. L. (2002). Aroma active components in aqueous kiwi fruit essence and kiwi fruit puree by GC–MS and multidimensional GC/GC–O. Journal of Agricultural and Food Chemistry, 50, 5386–5390. Joubert, E., Joubert, M. E., Bester, C., De Beer, D., & De Lange, J. H. (2011). Honeybush (Cyclopia spp.): from local cottage industry to global markets — the catalytic and supporting role of research. South African Journal of Botany, 77, 887–907. Koch, I. S., Muller, M., Joubert, E., Van der Rijst, M., & Næs, T. (2012). Sensory characterization of rooibos tea and the development of a rooibos sensory wheel and lexicon. Food Research International, 46, 217–228. Lawless, H. T., & Heymann, H. (2010). Descriptive analysis. Sensory evaluation of food, principles and practices (2nd ed.). New York, USA: Springer.

22

K.A. Theron et al. / Food Research International 66 (2014) 12–22

Le Roux, M., Cronje, J. C., Burger, B. V., & Joubert, E. (2012). Characterization of volatiles and aroma-active compounds in honeybush (Cyclopia subternata) by GC–MS and GC–O analysis. Journal of Agricultural and Food Chemistry, 60, 2657–2664. Le Roux, M., Cronje, J. C., Joubert, E., & Burger, B. V. (2008). Chemical characterization of the constituents of the aroma of honeybush, Cyclopia genistoides. South African Journal of Botany, 74, 139–143. Naudé, Y., & Rohwer, E. R. (2013). Investigating the coffee flavour in South African Pinotage wine using novel offline olfactometry and comprehensive gas chromatography with time of flight mass spectrometry. Journal of Chromatography A, 1271, 178–180. Őműr-Őzbek, P., & Dietrich, A. M. (2008). Developing hexanal as an odor reference standard for sensory analysis of drinking water. Water Research, 42, 2598–2604. Piccino, S., Boulanger, R., Descroix, F., & Shum Cheong Sing, A. (2014). Aromatic composition and potent odorants of the “specialty coffee” brew “Bourbon Pointu” correlated to its three trade classifications. Food Research International, 61, 264–271. Pino, J. A. (2014). Odour-active compounds in papaya fruit cv. Red Maradol. Food Chemistry, 146, 120–126.

Ross, C. F. (2009). Olfaction. In S. Clark, M. Costello, M. Drake, & F. Bodyfelt (Eds.), In: The sensory evaluation of dairy products (pp. 23–28) (2nd ed.). New York, USA: Springer. Ryan, D., Prenzler, P. D., Saliba, A. J., & Scollary, G. R. (2008). The significance of low impact odorants in global odour perception. Trends in Food Science & Technology, 19, 383–389. Schieberle, P., & Schuh, C. (2006). Aroma compounds in black tea powders of different origins — changes induced by preparation of the infusion. Developments in Food Science, 43, 151–156. Shapiro, S. S., & Wilk, M. B. (1965). An analysis of variance test for normality (complete samples). Biometrika, 52, 591–611. Stone, H., & Sidel, J. L. (1985). Descriptive analysis. Sensory evaluation practices (pp. 194–226). Orlando, Florida, USA: Academic Press. Vannier, A., Brun, O. X., & Feinberg, M. H. (1999). Application of sensory analysis to champagne wine characterization and discrimination. Food Quality and Preference, 10, 101–107. Wolters, C. J., & Allchurch, E. M. (1994). Effect of training procedure on the performance of descriptive panels. Food Quality and Preference, 5, 203–214.