Characterizing aromatic typicality of Riesling wines: merging volatile compositional and sensory aspects Armin Schüttler a,d,e, ⁎, Matthias Friedel c , Rainer Jung b , Doris Rauhut a , Philippe Darriet d,e a Hochschule Geisenheim University, Center of Analytical Chemistry & Microbiology, Department of Microbiology & Biochemistry, Von-Lade-Str. 1, 65366 Geisenheim, Germany b Hochschule Geisenheim University, Center of Wine Science & Beverage Processing, Department of Enology, Modeling & Systems Analysis, Von-Lade-Str. 1, 65366 Geisenheim, Germany c Hochschule Geisenheim University, Center of Viticulture & Horticulture, Department of General & Organic Viticulture, Blaubachstraße 19, 65366 Geisenheim, Germany d Univ. de Bordeaux, Institut des Sciences de la Vigne et du Vin, Unité de recherche Œnologie, EA 4577, 210 Chemin de Leysotte, 33882 Villenave d’Ornon, France e INRA, Institut des Sciences de la Vigne et du Vin, USC 1366 Oenologie, 33882 Villenave d’Ornon, France abstract article info Article history: Received 18 June 2014 Accepted 14 December 2014 Available online 20 December 2014 Keywords: Riesling aroma Sensory Typicality 3-Sulfanylhexan-1-ol (3-SH) Linalool 1,1,6-Trimethyl-1,2-dihydronaphthalene (TDN) Frequented free comment profiling (FFCP) Canonical correspondence analysis (CCA) The aim of this study was to characterize the perceived aromatic typicality of Riesling wines, the descriptors involved, as well as the chemical composition which leads to typicality perception. In total, 30 wines were tasted by wine experts and rated for Riesling wine aroma typicality. Then, descriptive analysis was undertak- en by a Frequented Free Comment Profiling (FFCP) approach permitting the judges to use their own vocabulary. In order to link sensory data to volatile chemical composition, 43 volatile compounds, including acetate and ethyl esters, high volatile sulfur compounds, polyfunctional thiols, monoterpenes, and C 13 -norisoprenoids were analyzed by gas chromatography techniques. Analysis of the results included the clustering of wines according to their Riesling wine typicality, which led to significant differences between all groups (p b 0.0001). Significant differences were also observed between wines for citation frequencies of several descriptor groups, including fruity (p b 0.001), minerality (p b 0.01), and vegetal odors (p b 0.001) as well as for several aroma compounds concentrations, including monoterpenes like linalool (b 0.5–147 μg/L; p b 0.042), α- terpineol (8–124 μg/L; p b 0.018), trans-linalool oxide (b 0.5–83 μg/L; p b 0.004) and a potent polyfunctional thiol, presenting grapefruit nuances, 3-sulfanylhexan-1-ol (3-SH) (218–1898 ng/L; p b 0.019). In order to relate analytical to sensory data more comprehensively, a Canonical Correspondence Analysis (CCA) was performed after variable selection using Partial Least Squares Regression (PLSR). This original approach proved to be a convenient tool for evaluation of mixed sensory and aroma compositional data sets consisting of parametric and metric data. Our results showed that higher concentrations of 3-sulfanylhexan-1-ol tended to contribute to perceived Riesling wine typicality, whereas no positive correlation was shown between linalool, recalling flowery nuances, and Riesling wines typicality. © 2014 Elsevier Ltd. All rights reserved. 1. Introduction One key dimension of wine quality is its ‘typicality’, which is related to its origin and the variety it is produced from (Charters & Pettigrew, 2007). Regarding the sensory appreciation of typicality, several related studies are based on cognitive psychological considerations of family resemblance (Rosch & Mervis, 1975) and inner categorical gradients of representativeness (Loken & Ward, 1990) by using a direct typicality measurement (Ballester, Dacremont, Fur, & Etiévant, 2005; Parr, Green, White, & Sherlock, 2007; Pineau, Barbe, Van Leeuwen, Dubourdieu, & Darriet, 2010). Overall, the global strategy includes sorting tasks for the identification of the most typical samples. These sorting tasks can then be complemented by descriptive tasks in order to associate sensory descriptors with typical and non typical wines. For this purpose, Descriptor Citation Frequency (DCF) and free comment methods approaches, in the form of a Frequented Free Comment Profiling (FFCP), have been shown to be convenient alternatives to conventional descriptive characterization as Quantitative Descriptive Analysis (QDA) in the field of sensory profiling of wines (Campo, Ballester, Langlois, Dacremont, & Valentin, 2010; Lawrence et al., 2013; Perrin et al., 2007). After characterization of the sensory properties of typicality, knowl- edge about the link between sensory perception of typicality and wine chemical composition (e.g. volatile compounds) is crucial to enable the control and management of typicality and therefore the quality of a wine. Generally, different multivariate methodologies can be used for relating sensory to compositional data, such as Principal Compo- nents Analysis (PCA) (Cliff & Dever, 1996), Correspondence Analysis (CA) (McEwan & Schlich, 1991), and Partial Least Squares Regression (PLSR) methods (Peng, Wen, Tao, & Lan, 2013; Tenenhaus, Pagés, Food Research International 69 (2015) 26–37 ⁎ Corresponding author at: Hochschule Geisenheim University, Center of Analytical Chemistry & Microbiology, Department of Microbiology & Biochemistry, Von-Lade-Str. 1, 65366 Geisenheim, Germany. Tel.: +49 6722 502 342. E-mail address: armin.schuettler@hs-gm.de (A. Schüttler). http://dx.doi.org/10.1016/j.foodres.2014.12.010 0963-9969/© 2014 Elsevier Ltd. All rights reserved. Contents lists available at ScienceDirect Food Research International journal homepage: www.elsevier.com/locate/foodres