Statistical Modeling of Tomato Firmness Evolution During Postharvest Storage

Abstract

In the quality assessment of horticultural products, a shift from destructive towardsnondestructive quality sensors can be observed. These nondestructive sensors offer the appealingpossibility to monitor quality changes of individual products over time, which was not possible before.Inherent to this shift, also an important change in experimental setting occurs since not allmeasurements are independent of each other. For instance, tomatoes with a low firmness today arelikely to have a low firmness tomorrow, too. Together with the knowledge that biological material ingeneral possesses a large variability, this type of repeated measures data necessitates a properstatistical analysis, and standard techniques used throughout literature in case of destructive sensorsare not applicable anymore. Therefore, this contribution aims at introducing the concept of mixedmodels in the field of postharvest technology. The possibility of fitting mixed models in commercialsoftware is rather recent (second half of the 90s) but has proven extremely useful in a broad rangeof application fields. In a practical experiment where the firmness of 13 different cultivars oftomatoes are monitored during postharvest storage with a commercial nondestructive firmnesssensor for two harvests, it will be shown that the mixed model concept allows quantifying intratomatovariability, inter-tomato variability and inter-cultivar variability, not only at a certain time point,but even as a function of storage time. The quantification of these different sources of variability is ofgreat importance to researchers and growers, but was up to now never accounted for.