Sensory analyses

The protocols of sensory tests were previously approved by the Ethics Research Committee of the State University of Ceará (Protocol No. 147,279). Before every experiment, informed consent was obtained from each panelist.

A trained panel, using descriptive sensory profiling (Stone and Sidel 2004), evaluated the tropical fruit juice blends. The panel consisted of 12 assessors screened for their sensory ability by basic tastes and odor detection tests as well as their ability to describe and discriminate food products. The generation of descriptors followed the Grid Method using samples of F1 formulation prepared in the laboratory, before and after pasteurization. During eight working sessions, the panelists, under the supervision of a moderator selected the descriptors that characterized the samples and established their definitions and references (Table 2). Panel validation consisted of the evaluation of three samples of F1 formulation: fresh, pasteurized, and overcooked. Descriptors were then quantified with a 9 cm non-structured linear scale, anchored on the extremes with terms that expressed their intensities. For each sample, the results were subjected to ANOVA, with two sources of variation viz. samples and repetitions, in order to check their discriminatory ability, repeatability, and consensus with the group.

Sensory descriptors, definitions and references used in descriptive analysis of tropical fruit blends

Subsequently, the two formulations of mixed juices (F1 and F2), before and after pasteurization, were evaluated in triplicate, following the experimental design of balanced complete blocks. Approximately 40 mL of juice was served at room temperature in glass cups covered with watch glass and coded with three-digit random numbers. Water was provided to cleanse the palate, and remove the residue of the juice. Tests were performed in individual acclimatized booths (22 °C) under controlled lighting (white day light, fluorescent). The booths were equipped with computer terminals for automatic data collection/retrieval of responses using FIZZ software version 2.40A.

Acceptance test was carried out with the tropical juice blends under the same conditions described above, with a difference that the juices were served cold (16 °C). The panel consisted of 50 untrained individuals, who were either employees or students from Embrapa Tropical Agroindustry. Panelists were regular juice consumers and non-smokers. Sixty percent of the panelists were women with 90% of the total panelists aged 20–35 years. Overall acceptability was calculated using a nine-point hedonic structured scale where 9 was designated as ‘like extremely’ and 1 was designated as ‘dislike extremely’ (Meilgaard et al. 2006).

As the storage period was too long to keep the panel trained, a Check-All-That-Apply (CATA) test was performed to evaluate the storage effect. A list with the descriptive terms generated in QDA plus two terms related to the color of the juices (red and brown) was provided and the consumers were asked to check all the terms that applied to the given sample.