Breeding tomato flavor : modeling consumer preferences of tomato landraces

Abstract

Tomato landraces are highly appreciated by consumers, who are willing to pay a price premium for them. But little is known regarding sensory perception and its relationship with fruit composition. The Spanish variety “Moruno” was selected as a model for this purpose. A collection of 30 populations was grown in different environments and evaluated by a consumer panel. Partial least square (PLS) models were then developed relating determinant flavor descriptors (sweetness, sourness, taste and aroma intensity, aftertaste persistence and agreeability, and overall flavor acceptability) with compositional variables such as soluble solids, pH, titratable acidity, individual sugars, organic acids, volatiles, and derived variables. PLS models identified relationships that had not been uncovered with correlation and simple regression analysis and offered low cross-validation errors (<15% of the range of variation). Although plant yield negatively affected sensory perception, it was possible to identify populations with a good combination of both traits.