Submitted to: Journal of Cereal Science
Publication Type: Peer reviewed journal
Publication Acceptance Date: 12/10/2007
Publication Date: 9/8/2008
Citation: Kepiro, J.L., McClung, A.M., Chen, M., Yeater, K.M., Fjellstrom, R.G. 2008. Mapping QTL for milling yield and grain characteristics in a tropical japonica long grain cross. Journal of Cereal Science 48:477-485. Interpretive Summary: Milling yield, also called "head rice recovery", is defined as the percentage of head rice, or whole kernel milled rice, obtained from rough rice (paddy rice) after milling. Breeding for improved milling yield is difficult because the trait exhibits a complex inheritance. Milling is a multi-step process that includes hulling, milling and separating, which gives rise to three components: brown rice (BR) recovery, milled rice (MR) recovery, and head rice (HR) recovery, respectively. Each milling component may be affected by multiple traits (sub-components) like kernel dimension, kernel hardness, bran thickness, etc. Many of the sub-component traits are under the control of numerous genes, and therefore milling yield and its sub-components have quantitative inheritance. Furthermore, these sub-component traits may be affected by non-genetic factors, including cultural management, weather conditions prior to harvest, and post-harvest handling of the grain. This makes evaluation very labor intensive and selection of superior milling varieties very difficult. The objective of this study was to identify molecular markers linked to milling components and their sub-component traits that will give breeders new tools to efficiently select genotypes with superior milling. A mapping population was developed using Cypress and Panda, two cultivars that are well adapted for production in the southern US and have been used in development of other commercial cultivars. Cypress is widely recognized as having high and stable milling yield (~64%) over a range of harvest moistures whereas Panda is characterized as having low milling yield (~ 52%). Both parents are early maturing long grain tropical japonica cultivars which reduce the confounding effects due to grain shape and days to maturity. Innovative phenotyping methodologies utilizing a WinSeedle™ scanner allowed accurate measurements of large numbers of seeds for grain dimension traits and objective pixel color classification for assessing chalk and green kernels. We identified robust markers for grain dimensions and the strong correlations between years for these traits indicates that these grain dimension traits are quite stable across environments and can be selected for with markers to make rapid progress. This may encourage the use of diverse germplasm which may have grain dimensions that do not fit within the long grain market class but have breeding value for other agronomic traits (yield and pest resistance). We also identified four QTL associated with the milling components of brown rice, total milled rice, and head rice recovery, each explaining 13-20% of the phenotypic variance. It was found that breakage of the grain following hulling could be used as a predictor of head rice yield and could be used to reduce large breeding population to a more manageable size before initiating the labor intensive process of milling. For most of the sub-component traits we found more than one QTL, and that each QTL typically explains between 10% and 20% of the variance in the phenotype measured. Next steps in our research will be to create regression models to identify the milling sub-components having the largest impact on milling yield. The DNA markers identified in the present study can then be used to select for the most important sub-component traits to increase milling quality.
Technical Abstract: Percent whole milled grains or milling yield is an economically important trait of commercial rice (Oryza sativa L.) because it largely determines the price per bushel that farmers receive for their crop. To investigate the inheritance of milling yield, a long grain japonica mapping population segregating for milling yield, grain quality, grain appearance, and agronomic traits was evaluated over two years at one location. A linkage map was constructed with 532 amplified fragment length polymorphism (AFLP) markers to provide genome wide coverage and 39 simple sequence repeat (SSR) markers were used to anchor the AFLPs onto the rice genetic map. Quantitative trait loci (QTL) analysis was conducted using 34 trait variables that produced 69 QTLs at 19 chromosomal regions. One QTL explaining 20% of the variance in brown rice (BR) recovery; two QTLs for milled rice (MR) recovery explaining 14% and 13% of the variance; and one QTL for head rice (HR) recovery explaining 14% of the variance, were detected. The QTL associated with HR was co-located with eight other traits affecting grain milling quality, cooking quality, and grain appearance, indicating the large impact this region has on rice grain quality. In this same region, we identified a QTL associated with Pre-Broken (PB); brown rice that is broken after rough rice is shelled but before it is milled. PB was highly correlated to HR (r = -0.78), which suggests that PB could be used as a predictor of head rice yield before initiating the labor intensive process of milling. An image analysis system was used to quantify chalky and green, immature grains, factors which impact milling quality. Two QTLs for chalk in brown rice, and one in head rice were identified as well as one QTL explaining 22% of the variance in green area of kernels. Four QTLs for kernel length of brown rice were detected explaining a cumulative variance of 68% and 58% in each of the two years that the study was conducted. Only one of the kernel length QTLs was detected in (milled) head rice; however, it explained 25% of the variance. Of the four QTLs identified for kernel width in head rice, the largest, explaining 19% of the variance was also detected in brown rice. These results demonstrate that kernel length and width are stable over growing environments and markers associated with these traits can be used as effective selection criteria. The identification of QTL explaining 13-20% of the variance in milling yield components (BR, MR, and HR, PB) as well as other QTL associated with grain characteristics important to rice grain quality will increase the effectiveness of rice breeding efforts to improve the milling quality and economic value of rice cultivars through marker assisted selection.