Résumé
To be able to account for sensory qualities earlier in the assessment of a new banana hybrid in a selection scheme, predicting the sensory perception of banana texture and taste by instrumental parameters was investigated. Thirteen cultivated banana and four new triploid hybrids were characterized by sensory profiling, and rheological and chemical analyses. Multilinear regressions were used to calibrate predictions using 13 cultivated bananas, and the quality of predictions was validated using four hybrids. The sensory characteristics sourness and sweetness were predicted by titratable acidity (R-2 = 0.68) and pH (R-2 = 0.66). Malate and citrate were the main contributors to sweetness and sourness. Astringency was predicted by total tannins (R-2 = 0.55). Rheological parameters from texture profile analyses (stress at fracture, fracturability) were more suitable than pulp puncture force to predict the sensory texture properties firmness (R-2 = 0.47) and melting (R-2 = 0.60). These textural properties were predicted by titratable acidity and dry matter content (R-2 = 0.62). Predictions of mealiness, adhesiveness, and heterogeneity were not efficient. Differences of 3.6-3.7 meq 100 g(-1) FW in titratable acidity or of 0.30 g 100 g(-1) FW in malate or. citrate were required to ensure a detectable difference in sourness or sweetness (p = 0.9). Pulp puncture force needed to differ by a minimum of 0.9 N before a difference in firmness could be perceived by the panelists. In conclusion, while models to predict sourness and sweetness can now be used for high throughput phenotyping, we recommend additiqnal tests for other sensory attributes. (C) 2013 Elsevier B.V. All rights reserved.