Résumé
Conventional wheat grain milling processes aim at dissociating and isolating the starchyendosperm, recovered in flour or semolina, from the outer layers as bran fractions.Proportions of each tissue in mill streams impact their nutritional quality or their end-useproperties. Among various methodologies, measurement of biochemical markers specificallylocated in the different wheat grain tissues was successfully applied to evaluate these tissueproportions within fractions. However this method implies at least four differentmeasurements and, due to variability in marker amounts, could lead to some loss of accuracy.Considering the diversity in neutral ose composition between each grain tissue, this data,coming only from one biochemical analysis, could be useful to quantify the tissue proportion.The objective of this study was first to evaluate the specificity and variability of neutral osecomposition in durum and common wheat grain tissues and then to evaluate its ability topredict tissue composition in various mill streams based on this characterization. Pure starchyendosperm, aleurone layer, outer pericarp were dissected as well as a composite layer made ofnucellar epidermis+testa+inner pericarp. Starchy endosperm, with high amount of glucose(related to the presence of starch) was easily differentiated from peripheral tissues (richer inarabinose and xylose present in wheat cell wall polysaccharides). Based on measured tissueproportion obtained by another method, the ose composition in mill streams was predictedfrom pure tissue composition: the best results were obtained for arabinose and xylose contentbut with a global error of 13 %. On the other hand, multiple regression methods (PLS, MCR)were used to predict tissue proportion in mill streams based on ose composition. If outerpericarp and starchy endosperm could be predicted with a satisfactory error of 5%, predictionof the aleurone layer and intermediate layer containing testa was not straightforward witherror higher than 10%. The high similarity between these tissues could be part of theexplanation. Therefore even if variability in ose composition due to different cultivars was notfound significant, the specificity for each tissue appeared insufficient to envision suchanalyses to determine the tissue composition in mill streams.