SENSORY PROFILE CHARACTERIZATION AND TYPICALITY ASSESSMENT OF PDO “BAIRRADA” AND PGI “BEIRA ATLÂNTICO” RED WINES CARACTERIZAÇÃO DO PERFIL SENSORIAL E AVALIAÇÃO DA TIPICIDADE DOS VINHOS TINTOS DOP “BAIRRADA” E IGP “BEIRA ATLÂNTICO”

The distinctive Portuguese wines from Beira Atlântico region, encompassing the protected designation of origin (PDO) Bairrada and the protected geographical indication (PGI) Beira Atlântico, were investigated by a wine expert panel through descriptive analysis and through assessment of typicality. For that 19 trained tasters performed a blind sensory evaluation of 21 representative wines from those designations. The variables considered were the color tonality and color intensity, aroma intensity, 18 aroma descriptors and 14 taste descriptors. Typicality was investigated through a single question, where the assessor was asked to score if the sample is a good or bad example of the type. Of the 21 wines selected, seven were classified as PGI or Regional, eight as PDO and six as “Clássico”. Firstly, differences were analyzed between wine types considering all variables followed by clusters analyses confirmation. We could not find any difference between Regional and PDO Bairrada wines in terms of typicality and sensory profile. However the small group of “Clássico” wine was clearly identified by the tasters as being more typical, with also significant differences on sensory evaluation. Secondly, centered means analysis (CMA) of the 18 aroma and 14 taste items were performed to identify which of them are considered to be more distinctive. Thirdly, an exploratory factor analyses (EFA) by the principal component method (PCM) was applied to data, allowing identification of five vectors which aggregate the aroma items and four vectors which aggregate taste items. Finally, data collected from a sample of 20 questionnaires from a previous study based on cognitive knowledge and long term memory of 20 wine experts interviewed over the same 18 aroma variables and 14 taste variables, was analyzed under the same principal components (PC) and compared. The 21 wines representatives of PGI Beira Atlântico and PDO Bairrada can be defined as being medium-high intensity, ruby colored wines, having a woody & spice, ripe fruit aroma profile with also herbal and mineral aromas. On taste they tend to have a pronounced component of acidity & astringency, balanced with a smooth & sweet taste component, with a very persistent finish.


SUMMARY
The distinctive Portuguese wines from Beira Atlântico region, encompassing the protected designation of origin (PDO) Bairrada and the protected geographical indication (PGI) Beira Atlântico, were investigated by a wine expert panel through descriptive analysis and through assessment of typicality.For that 19 trained tasters performed a blind sensory evaluation of 21 representative wines from those designations.The variables considered were the color tonality and color intensity, aroma intensity, 18 aroma descriptors and 14 taste descriptors.Typicality was investigated through a single question, where the assessor was asked to score if the sample is a good or bad example of the type.Of the 21 wines selected, seven were classified as PGI or Regional, eight as PDO and six as "Clássico".Firstly, differences were analyzed between wine types considering all variables followed by clusters analyses confirmation.We could not find any difference between Regional and PDO Bairrada wines in terms of typicality and sensory profile.However the small group of "Clássico" wine was clearly identified by the tasters as being more typical, with also significant differences on sensory evaluation.Secondly, centered means analysis (CMA) of the 18 aroma and 14 taste items were performed to identify which of them are considered to be more distinctive.Thirdly, an exploratory factor analyses (EFA) by the principal component method (PCM) was applied to data, allowing identification of five vectors which aggregate the aroma items and four vectors which aggregate taste items.Finally, data collected from a sample of 20 questionnaires from a previous study based on cognitive knowledge and long term memory of 20 wine experts interviewed over the same 18 aroma variables and 14 taste variables, was analyzed under the same principal components (PC) and compared.The 21 wines representatives of PGI Beira Atlântico and PDO Bairrada can be defined as being medium-high intensity, ruby colored wines, having a woody & spice, ripe fruit aroma profile with also herbal and mineral aromas.On taste they tend to have a pronounced component of acidity & astringency, balanced with a smooth & sweet taste component, with a very persistent finish.

INTRODUCTION
Sensory characterization of food and beverages has amongst its aims detection of unique or distinguishable qualities in a product, which can be influenced by the place of origin (Green et al., 2011).Wine is one product whose concept of typicality is prevalent (Maitre et al., 2010) and the concepts of terroir, protected origin, and typicality incorporate the notion that wines from delimited geographical regions can be perceived as different (Parr, 2009).A wine is typical when some of its characteristics, which reflect both its origin and terroir, can be identified and make it recognizable as belonging to a distinctive type (Charters and Pettigrew, 2007;Maitre et al., 2010).Therefore, typicality includes sensory, technical, and environmental dimensions (Cadot et al., 2010).
According to Mueller et al. (2010), product expectations at the initial purchase and intrinsic sensory attributes during product consumption influence the repurchase decision.Thus, when consumers choose a wine bottle based on its identity or terroir, it is because they believe that there is a combination of sensory attributes that differentiates that wine from the others.Typicality and quality used to be correlated and are based on intrinsic cues such as pleasure and appearance, gustatory, and olfactory characteristics (Charters and Pettigrew, 2007).Applying the sensory approach to quality control within the food industry of typical products is fundamental for preserving the market position, maintaining consumer confidence and loyalty toward the product (Endrizzi et al., 2013).It does not mean that a typical wine is always good, but the concept could reduce the risk of an incorrect purchase.If the expectation of the wine is achieved, the possibility of repurchase increases, conferring an added commercial value.For this reason, recognition of typicality has become an economic factor for wine regions and winemakers.
Some studies have reported that wine experts learn to categorize wines according to the variety and are able to recognize them blind, while the identification of origin is less successful (Ballester et al., 2008).In the investigation of protected designation of origin (PDO) typicality is not straightforward, as it seems that there is more sensory overlapping between close PDO than between wines made with different grape varieties.The existence of a defined and distinguished sensory space in a PDO is not obvious and the results depend on the samples choice, the professional tasters' selection and the PDO itself (Maitre et al., 2010).There have been numerous studies characterizing regional sensory differences in wines (Kontkanen et al., 2005;Perrin et al., 2008;González-Álvarez et al., 2011;Green et al., 2011;Cadot et al., 2012;Robinson et al., 2012;Johnson et al., 2013).Some other studies have compared the sensory profiles of wines from multiple countries, including red and white wines from Ibero-American countries (Tonietto et al., 2014), red wines from South American countries (Llobodanin et al., 2014), red wines from Australia and China (Williamson et al., 2012), and Sauvignon Blanc wines from France and New Zealand (Parr et al., 2010) and either Austria (Green et al., 2011) or South Africa (Lund et al., 2009) or Argentina and California (King et al., 2014).
In Portugal, few studies have been attempted to assess for typicality or characterize wine regions profiles.Jose-Coutinho et al. (2012, 2013, 2015a, 2015b) investigated white and red wine sensory profiles of the 12 protected geographical indications of mainland Portugal.Evidence of clusters grouping several wine regions under the same sensory profile suggests that these 12 regions could be grouped into three to five major macro zonings that originate a much broader territorial reading.The authors also highlighted the benefits associated to this aggregation in a small country such as Portugal, encompassing low yield viticulture and dominance of old vines in small parcels owned by even smaller producers, and suggested that scale factors might result in significant gains in areas such as wine certification, marketing and communication.
A large number of sensory methodologies have been used with professional tasters to show the sensory characteristics of a wine category and several papers deal with the question of how typicality can be measured and how different methods have been suggested to evaluate wine typicality (Maitre et al., 2010).Sensory methods employed in the current study included analytical evaluation by descriptor ratings of intensity of selected aromas and flavors, with intensity considered to be correlated with perception of aroma and flavor concentration.For typicality assessment we also employed a variant of the sensory methodology developed from Ballester (2004) where the author suggested investigating typicality with a single question, where the assessor is asked to score if the sample is a good example or not of the type.
The aim of the present study was to identify a sensory profile for red wines from Beira Atlântico region and to evaluate sensory characteristics and typicality differences between three distinct types of wines from this region, known as PGI wines, PDO wines and "Clássico" wines.We also investigated whether there is an agreed ideal for typicality between wine professionals, and identified the specific aroma and taste items that are essential and contribute for wine typicality; that is, making a given wine a good example of the concept.

Wines
The wines studied were commercially available red wines produced in PDO Bairrada and PGI Beira Atlântico.The main differences between these two designations are that the area of Beira Atlântico is noticeably larger than PDO Bairrada, allowing higher yields and a broader range of varieties.According to the Portaria nº 212/2014, within PDO Bairrada it is possible to apply for the mention «Clássico» ensuring that the wine meets a number of requirements such as: minimum of 30 months of aging which 12 of them in bottle, minimum 12.5% of alcohol, Baga variety in a minimum of 50% in lot composition and a maximum yield of 55 hL per ha.21 wines were selected in collaboration with Bairrada certification authority to represent the diversity of three type of wines produced in these regions, from vintages 2009, 2010 and 2011, with retail prices ranging from 1,29€ to 29,95€ per bottle of 750 mL.All wines were vinified by classical red winemaking technology with extended maceration and fermentation.The best-selling wines for each type of wine were selected, making the sampling representative of the region.Of the 21 wines, seven were classified as Protected Geographical Indication (PGI) or Regional, eight as Protected Designation of Origin (PDO) and six as "Clássico".These six wines, selected to integrate "Clássico" group, had the necessary characteristics to be categorized as «Clássico» wines (although some of them were not bottled and labeled as such).They share some common characteristics such as: coming from old vineyards, minimum 30 months of aging which 12 of them in bottle, minimum 12.5% of alcohol, Baga variety in a minimum of 85% in lot composition and a maximum yield of 55 hL per ha.

Participants
The sensory expert panel was composed of 19 judges related to wine industry (winemakers, wine brokers, sommeliers and academic oenology professors), and was selected on the basis of their experience in wine tasting, sensory performances, interest and training to perform wine descriptive analysis (DA).No specific training was provided for participants immediately prior to their participation in the current study.All participants had an extensive experience of PDO Bairrada and PGI Beira Atlântico wines and most of them already integrate the panel tasting of Bairrada certification authority, which has its own methodology to control the consistency of their tasters' evaluations.

Design and Procedure
The study was fully within-subject and was conducted in three one-hour sessions on the same morning day separated by a 15-minute break, at Estação Vitivinícola da Bairrada (Anadia).Each session contained seven wine samples codified with a random alphanumeric code, with the order of samples randomized in each session.
Wine samples were stored and presented at 20 ºC for detection of color, aroma and taste.The bottles were opened immediately before the analysis and panelists were prevented from seeing their label or shape.Samples were presented in standard wine tasting glasses (ISO 3591).Panelists were informed that they were evaluating wines from PGI Beira Atlântico and PDO Bairrada, including «Clássico» wines.However, for each wine sample no specific information was provided concerning the grape varieties, type or price.
The environment for tasting was controlled as advised for sensory laboratories and international wine competitions.There was a uniform source of lighting, absence of noise and distracting stimuli, and ambient temperature was around 22ºC.Participants were then seated in separate booths and water was available to clean the palate.

Questionnaire
The questionnaire was designed according to the approved International Organization of Vine and Wine method for sensory assessment of wines, thus it is divided in three blocks: visual sensations (Color), the aromatic sensations (Aroma) and the gustatory and tactile sensations (Taste).Figure 1 shows an extract of the questionnaire.Extrato do questionário de análise sensorial.
DA was carried out and the assessors scored the intensity of each attribute using a 10 point structured and discontinuous scale.Scale were anchored with the terms "low intensity" on the left and "high intensity" on the right.
The color was assessed by its intensity and by its tonality, with color patterns of red wines according to the latest studies (Hernández et al., 2009;Jose-Coutinho et al., 2012, 2013, 2015b), which is evaluated for the tone violet-purple; purple-ruby; ruby-garnet or garnet-brick red and for the intensity in a 0-10 rating scale.
Aroma was evaluated by one overall measure for intensity and 18 aroma descriptors, which tended to summarize significant amount of scientific research, many of which already used in the form of aroma wheels (Fischer et al., 1999).
The taste of red wines, measured via tactile and gustatory sensations by the panel of experts, was evaluated under 14 descriptors, including the classical essential tastes (sour, sweet, salty, bitter), and adding the result of recent work focused on the tactile sensations resulting from various types of bitterness and astringency and the sensations caused by caustic and drying alcohol sensations (Gawel et al., 2000(Gawel et al., , 2001)).The wine texture was evaluated in a tripartite form, considering smooth, rough and grain sensations on mouthfeel (Jose-Coutinho et al., 2015b).
The same 19 judges also assessed the wine typicality.
For each wine, the assessors were instructed to answer the question "Do you think this wine is a good or bad example of what a Bairrada wine is?"These instructions were derived from Ballester (2004).The assessors were asked to rate the wines sample's typicality on a structured and discontinuous scale, 0-10 rating, anchored at left by "bad example" and at right "good example".

Data analysis
Data acquisition was assisted by SPSS software (IBM Statistics SPSS Version 20).Over 14.000 database entries were retrieved, placed in an Excel worksheet and subsequently analyzed using SPSS software.The effect of the tasters was not considered and the average was computed for each item.
Firstly, a descriptive statistics was performed in order to analyze various sensory descriptors for the three types of wines.Secondly, a group of inferential tests was applied to look for significant characteristics.One way ANOVA, with Bonferroni test, was performed to test differences between wine types and evaluate statistically differences between them.An agglomerative hierarchical cluster analysis with square Euclidean distance, a tool that creates subgroups based on proximities of data (pair wise distance between observations) and inter-relationships of variables, was performed to better understand the interrelationships between the 21 wines.It classifies objects so that each object is similar to others in the cluster with respect to a predetermined selection criterion.The resulting clusters of objects should then exhibit high internal (within-cluster) homogeneity and high external (between-cluster) heterogeneity.
It should be highlighted that this is an exploratory study were a profile should be shaded for red wines of this region.This profile has three components: the aroma, the taste and the color.The typicality is also analyzed.It is aimed to find the principal characteristics of the wine based on components to be ease to compare with other regions in further studies.
For typicality assessment, the average typicality scores were achieved and ANOVA was performed.Spearman correlation analysis (Spearman's rank-order correlation), whose coefficient measures the strength of association between two ranked variables, was also applied to find out which sensory characteristics (for aroma, taste and color), are associated with typicality judgment.The Spearman correlation can be used with the assumptions of presence of a monotonic relationship between variables and they are either ordinal, interval or ratio.
A principal components analysis with Varimax rotation, sought to explore the possibility of reducing the initial set of sensory descriptors on a fewer dimensions -the main components.The number of retained components was based on Kaiser test (according to which the components with an eigenvalue grater or equal to one are retained).After reducing the initial space variables a sensory profile was identified.
The methods used try to measure which items are significant and which are correlated with the wines from this region.
Data from a previous study based on memory and cognitive knowledge of Beira Atlântico red wines, from 20 wine experts interviewed over the same 18 aroma variables and 14 taste variables (Jose-Coutinho et al., 2012, 2013, 2015b) was collected and analyzed under the same principal components (PC) and compared.Centered means were used to allow comparison between components avoiding the weight since components do not have the same number of items.
Using the components obtained, an ANOVA was performed to analyze the differences between wine types.

Comparing wine types
The three types of wine in study were analyzed and compared in terms of color intensity and tonality, aroma intensity, 18 aroma and 14 taste descriptors, and typicality.Because further analyzes were performed to aggregate aroma and taste descriptors into vectors, they were already analyzed and presented separately from the others.
Therefore, the first analysis comprises the variables of color, nose intensity and typicality.Table I shows the frequencies and the means for those variables indicating that the "Clássico" type was considered to be more color and nose intense and more typical.An ANOVA was performed to test differences between wine types and it showed that the "Clássico" type is statistically different from the others.The tonality most referred (mode) was "ruby-garnet" followed by "purple-ruby" consistent in all types.This ruby predominant color could be explained by the low pH wines (Glories, 1984), and therefore high natural acidity, which are common in Beira Atlântico region wines based on Baga variety (Böhm, 2007;Cardoso, 2007).Table II shows the results of aroma items for the three types of wine.It can be seen, after the ANOVA test, that "Clássico" type is statistically different from the other two on black fruit, raisin, jam, pastry, spice, caramel and woody presence, and different from the regional on mineral presence, having a higher level in all of those items.The results of the ANOVA test for taste are presented on Table III.It can be seen that the "Clássico" type is statistically different from the other two on grain texture, astringent, full body, alcohol, oily and length items and from the regional type on tactile dryness and rough texture items, having a higher presence level in all of them.A Clusters Analysis was performed in order to better understand the interrelationships between the wines that comprise the three types of wines.This methodology is useful to confirm if the types are consistent in terms of color, nose, taste and typicality.
It would be expected that the wines will be grouped according to the type they belong.The dendograms try to show if the wines are grouped according with their classification.Although the wine classification depends on a higher number of characteristics, it is important to see how the evaluation characterizes the wine; in that sense this analysis works as a convergent validation of the sensory analysis.Figure 2 show a tendency to aggregate wines from "Clássico" group, although with some wines from regional and PDO groups being mixed.This suggests that wines from "Clássico" group have contributed to discriminating from others by the panelists, especially in what respects to typicality. .Cluster analysis for "Typicality" (a) and for "All items" (b) considering all wines and types -Lowercase letters "r", "d" and "c" before each alphanumeric wine code, refers to wine type and means Regional, PDO and "Clássico", respectively.

Typicality assessment
Typicality was investigated through a single question, where the assessor was asked to score if the sample is a good or bad example of the type.
Table IV shows the significant correlations between typicality and either color, aroma and taste items, allowing to identify which sensory characteristics contribute to define typicality of Beira Atlântico red wines.The relationship between typicality and items from descriptive analysis showed that full body, oily and length are taste descriptors highly correlated with evaluation of wine typicality by the expert panel.We can also assume that aroma and color intensity are important items correlated with typicality judgment by the expert panel, being the ripe fruit (black fruit, stone fruit, jam), woody and spice the aroma descriptors.
These achievements help to better understand why tasters considered "Clássico" wines more typical and placed this group as a distinct type of wine.We could find that those descriptors identified as being tightly connected with typicality judgment strongly match with those identified on Tables I, II and III that had showed "Clássico" type being statistical different from Regional and PDO types.

Evaluating aroma and taste distinctive descriptors of PGI Beira Atlântico and PDO Bairrada wines
Figure 3 represents the centered means for the 18 aroma and 14 taste items in descending order.
It can be seen that woody, spice, herbal, black fruit, stone fruit, mineral, red fruit and jam are the aromas considered more distinctive of these wines.In what concerns to taste items, it can be seen that length, full body, astringent, oily, alcohol, acid, smooth texture and tactile dryness are the taste items considered more distinctive of these wines.
Data points were spread out over a large range of values, remembering how difficult is for the panelists to evaluate single sensory descriptors in a disaggregated form, leading us to perform principal component analysis (PCA).

Defining vectors which aggregate aroma and taste descriptors
An exploratory factorial analysis (EFA) was conducted, with the principal component method, separately for aroma and taste items.
Principal components factor analysis was carried out for the various sensory parameters that constituted the aroma characterization survey of red wines, in accordance with the most up-to-date bibliography and it was obtained five components explaining 60.77% of the variance.The factor loadings for aroma items are presented in Table V.
The results of this analysis suggest that several parameters can be grouped, since their variation is directly (or inversely) proportional, among them.Group 1 includes the following aroma descriptors: mineral; raisin; dried fruits; jam and pastry, which may be grouped under the so called "overripe" aroma profile.In a first glance it could appear strange the presence of the minerality item on this group, however this overripe aroma on wines, which means high levels of grape maturation, might be connected with deeper rooting systems embedded in the calcareous-clay soils present in the region, which could confer simultaneously this minerality character.
The second group joins the descriptors red fruit, black fruit and stone fruit, assigned to the "ripe fruit" profile.Group 3 gathers spice, caramel and woody easily recognized as members of the "woody & spice" profile.Group 4 joins floral, dried flowers, herbal and citrus fruit, all included in the "floral & fresh" profile.The fifth Group aggregates "green & off flavor" profile, including chemical, vegetal (or green) and animal.each component, respectively: 13.700; 12.736; 11.711; 11.372; 11.252.The EFA for taste shows four components explaining 62.87% of the variance.The factor loadings for taste items are presented on Table VI.Group 1 is composed by full body, alcohol, oily and length easily recognized as members of the "persistent" profile.The second Group joins the tastes sweet, smooth texture and (the inverted assessment corresponding to) tactile dryness, and may be referred to as "smooth & sweet".We must emphasize the opposite assessment between "tactile dryness" and the two descriptors "smooth texture" and "sweet" profile (associated with "smooth & sweet" profile) made by the expert's panel, because the panel discriminates them inversely.That is, when the panel of experts considers that a particular wine is characterized by having high levels of flavor descriptors "sweet" and "smooth texture", then the evaluation of "tactile dryness" in that wine is inversely proportional.Group 3 includes salt, bitter and bubbly to express the "bitter & salt" profile.Bitter taste was unexpectedly assessed differently from astringent sensations as they usually appear correlated (Lesschaeve and Noble, 2005) and its inclusion in a "bitter and salt" profile can be related with sensations of under maturation of the grapes.Group 4 adds grain and rough textures, astringent and acid taste, encompassing the evaluation of a rough sensations profile designated by "acid & astringent".This grouping of acid and astringent sensations can be explained by the well known influence of the acidity in the increasing astringency perception (Fontoin et al., 2008).

Aroma and taste radar profiles
Table VII presents the principal components and means for the data collected from the 399 questionnaires applied in the present research and for the data collected from questionnaires applied in Jose-Coutinho et al. (2012, 2013, 2015b) research, which was here analyzed under the same principal components.Both samples are showed but it must be noted that the first sample is based on a real sensory experience (and wine comparative) and the second is based on cognitive memory and expert knowledge (in absolute terms without any comparison).Because this experimental design leads to intensity differences, only presence or absence of the descriptors is analyzed.The methodology follows Jose-Coutinho et al. (2012, 2013, 2015b) work, and the components obtained grouped the items in a correct conceptual way. Figure 4 represents the centered mean aroma components for both samples and it can be seen that although graphic areas do not coincide exactly, one can argue that both vectors "woody & spice" and "ripe fruit" have a significant presence in the aroma profile of PDO Bairrada and PGI Beira Atlântico wines.These results corroborate findings on the free varietal and pre-fermentative related volatile compounds in crushed grapes from Baga cultivar (Coelho et al., 2006).In the cited research, the sesquiterpenoids was the group with higher number of constituent varietal compounds and reported as exhibiting spicy and woody aromas that could contribute favorably to the wine aroma characteristics.Another research analyzed the volatile composition of Baga single-varietal red wine.According to aroma index and odor descriptor, the nine compounds that seemed to be the most powerful odorants present in the wine are guaiacol, 3methylbutanoic acid, 4-ethoxycarbonyl-γbutyrolactone, isobutyric acid, 2-phenylethanol, γnonalactone, octanoic acid, ethyl octanoate and 4-(1 hydroxyethyl)-γ-butyrolactone).
It has been concluded that it presents ripe red fruit, cherry, strawberry, vegetal, sweet and smoke/phenolic notes (Rocha et al., 2004), which are also closely correlated with the aroma profile identified in our present research, suggesting that the Baga cultivar plays an important role on aroma profile of PDO Bairrada and PGI Beira Atlântico wines.The "green & off flavor" component in wine tasting is not present at same intensity level as it was in Jose-Coutinho et al. (2012, 2013, 2015b) research based on cognitive knowledge.Conversely, the "overripe" aroma component is now more evident.These facts could be related to improvements on viticultural and winemaking practices and also facilitated by warming climatic changes that have been encouraging grape ripeness, reducing its green character and inducing a shift to a more driven ripe fruit character.The temporal gap between conceptual and perceptual expertise (Cadot et al., 2012) can explain the differences achieved.
Thus, based on the information collected from principal component method and also considering the more distinctive aromas resulting from the centered mean analysis, we can argue that PDO Bairrada and PGI Beira Atlântico wines have a woody & spice, ripe fruit aroma profile with also herbal and mineral aromas.
The centered mean taste components for both samples are also represented in Figure 4.It can be seen that the wine profile for taste is mainly on the fourth quadrant "persistent" -"acid & astringent", but at same time balanced with a "smooth & sweet" taste component.The "bitter & salt" component is not present in a same intensity level.
Based on the information collected from principal component method and also considering the more distinctive taste items resulting from the centered mean analysis, we can argue that PDO Bairrada and PGI Beira Atlântico wines can be defined as having pronounced "acidity & astringency", balanced with a "smooth & sweet" taste component and being very persistent.

Analyzing wine type by principal components
Table VIII shows the aroma and taste components per wine type.It can be seen that the "Clássico" type is statistically different from the others on "overripe", "woody & spice", "acid & astringent" and "persistent", having higher presence in all of them.The center of interest of this study is the determinants of the aroma and the taste (evaluated on their own) thus a cluster analysis (Figure 5), considering all wines and types, was performed for the nine principal components of aroma and taste.Other characteristics and relationships may be addressed in future studies using the frame presented.We can see that the "Clássico" type tends to group together, although with some wines from regional and PDO groups being mixed, confirming a certain distinctiveness of this type of wine and also strength the adequacy of the components as the vectors of the PDO Bairrada and PGI Beira Atlântico wines.This tends to be a sensitivity analysis to the components, i.e., the components comply with the wine classification.

CONCLUSIONS
This is the first time that an extensive study has been attempted to characterize the sensory profile and typicality of PGI Beira Atlântico and PDO Bairrada wines.
These wines were characterized by having a ruby color, with medium-high intensity.On the nose they presented a medium to medium-high intensity, with a "woody & spice" and "ripe fruit" aroma profile with also herbal and mineral aromas.On taste, these wines can be defined as having pronounced "acidity & astringency", balanced with a "sweet & smooth" mouthfeel component and being very "persistent".
Results from this research, based on wine tasting and sensory experience, were compared with data obtained from Jose-Coutinho et al. (2012, 2013, 2015b) based on cognitive memory and expert knowledge.We can argue that methodology proposed by Jose-Coutinho et al. (2012, 2013, 2015a, 2015b) could be a valid alternative to the conventional wine tasting methods when intended to identify wine regions profiles, although further work on other regions might continue.
When comparing wine types, we could not find any difference between Regional and PDO Bairrada wines in terms of typicality and sensory profile, as Jose-Coutinho et al. (2012, 2013) had already achieved on his research.However, the small group of "Clássico" wine was clearly identified by the tasters as being more typical, with more color and nose intensity.In addition, it was also identified a substantial increase on its spice & woody character as well as more overripe fruit character on the nose.On the taste, "Clássico" wines tends to have a pronounced "acidity & astringency" component when compared to regional and PDO Bairrada wines and are also more persistent.These results point to "Clássico" group reached a distinct level and suggest that Bairrada region has an opportunity to differentiate these wines among others, keeping its own typicality and identity, facilitating its recognition abroad, which could lead to Bairrada affirmation in the global wine industry.However, when analyzing the mean prices of the three types of wines, we can realize that "Clássico" wines tend to be four and eight times more expensive than Regional wines and PDO wines, respectively, which could have a great impact on consumer choice.

Figure 1 .
Figure 1.Extract of the sensory questionnaire.
Figure 2. Cluster analysis for "Typicality" (a) and for "All items" (b) considering all wines and types -Lowercase letters "r", "d" and "c" before

Figure 3 .
Figure 3. Aroma (a) and taste (b) items obtained by centered means analysis and ordered by importance, considering all wines and types.Descritores do aroma (a) e do gosto (b), obtidos por análise de médias centradas e apresentados por ordem de importância, considerando todos os vinhos em prova.

Figure 5 .
Figure 5. Cluster analysis for the nine components of aroma and taste items considering all wines and types.Análise de clusters realizada para as nove componentes principais encontradas para os descritores do aroma e do gosto, considerando todos os vinhos em prova.

TABLE I
Frequencies and means related to the variables color, aroma intensity and typicality for the three types of wineFrequências e médias referentes às variáveis cor, intensidade aromática e tipicidade para os três tipos de vinho

TABLE II
Mean intensity of aroma descriptors for the three types of wine.

TABLE III
Mean intensity of taste descriptors for the three types of wine.

TABLE IV
Significant correlations between typicality and either color, aroma and taste itemsCorrelações significativas entre tipicidade e os descritores de cor, aroma e gosto

TABLE V
Aroma rotated component matrix with factor loadings for aroma items and their principal components Matriz dos descritores de aroma após rotação dos fatores com respetivas carga fatorial e componentes principais

TABLE VI
Taste rotated component matrix with factor loadings for taste items and their principal components Matriz dos descritores do gosto após rotação dos fatores com respetivas carga fatorial e componentes principais

TABLE VIII Principal
components means related to aroma and taste items for the 3 types of wine Médias de intensidade das componentes principais encontradas para os descritores de aroma e gosto, comparando os 3 tipos de