Differential QTL underlie wheat grain physical quality when measured using image‐based versus traditional laboratory methods

Abstract

BACKGROUNDThe marketing value of wheat (Triticum aestivum L.) is determined, in parts, by the grain's physical characteristics as they directly or indirectly influence milling performance and baking quality. These characteristics have been measured manually in the past, but now, digital image analysis (DIA) is increasingly being used to replace the slow phenotyping system. Here, we asked whether this could lead to the identification of the same or different genes, when compared to the traditional phenotyping methods.RESULTSWe measured grain physical quality on 142 wheat doubled haploids grown in the field over two years and using quantitative trait locus (QTL) mapping approach we found that (1) for wheat grain weight, the use of DIA provided genetic information that mostly conformed to those obtained using the traditional phenotyping methods, with heritability estimates that were identical across both methods. Majority of the QTL detected were consistent between the traditional versus digital phenotyping methods; (2) a more complex architecture, however, arose from QTL analyses of hectolitre weight (HLW) and percentage of shrivelled grains (SCR). The estimates for heritability varied by as much as 0.24 across methods and more significantly, many of the detected QTL for both traits were method‐specific; (3) though method‐specific, identified QTL mapped to genomic regions known to harbour genes for grain physical traits.CONCLUSIONSTGW is robust to phenotyping method, but a different genetic system underlie HLW and SCR, when these were measured using traditional versus digital image analysis. For these traits, heritability estimates were larger when phenotyped using traditional methods relative to digital image analysis, suggesting that further refinements are required to better correlate digital image analysis with the traditional phenotyping methods.This article is protected by copyright. All rights reserved.