Submitted to: Rice Technical Working Group Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: January 1, 2006
Publication Date: February 15, 2006
Citation: Kepiro, J.L., Fjellstrom, R.G., McClung, A.M. 2006. Molecular markers for milling yield in Southern U.S. long grain rice. Rice Technical Working Group Meeting, February 29-March 1, 2006, Houston, Texas. 2006 CDROM. Technical Abstract: Milling yield, defined as the percentage of whole rice kernels recovered after de-hulling rough rice, is a critically important trait in rice, with poor milling cultivars being rejected by rice growers. Milling yield is a complex trait comprised of component traits, each of which is under the control of numerous loci. Breeding for improved milling yield is difficult because of the numerous sub-component traits and their quantitative inheritance, and the impact of pre- and post-harvest environments on the grain. A population of 137 F11 progeny lines derived from a Cypress x Panda cross was developed for mapping quantitative trait loci (QTLs) associated with milling yield. Cypress is a long grain cultivar having intermediate amylose content and is well-known for high and stable milling yield (whole kernel percentage ~ 64%) over a wide range of harvest moisture levels. Panda is also a long grain cultivar, but has low amylose content, and is characterized by low milling yield (whole kernel percentage ~ 52%). Important component traits contributing to the final milling yield were identified by regression analysis. Subsequent QTL analysis was used to identify molecular markers linked directly to milling yield and/or to the component traits. Samples were milled using standard milling techniques beginning with 125g of rough rice. Grain recovered after each step of the processing was weighed and data for brown rice (BR), total milled rice (TR) and whole milled rice (WR) were recorded. The whole kernel recovery was determined as the proportion of whole milled grains to the total milled rice (WR/TR). Kernel lengths and widths were measured on 100 - 150 kernels per family for both brown and milled rice using a WinSeedle (2005a Pro) color image analysis system, and the broken and full length brown rice kernels were counted. The color analysis feature of the system allowed development of a new method for quantifying the area of chalkiness on a per kernel basis. Kernel thickness was measured with a digital micrometer on 20 kernels. The apparent amylose content was measured using standard procedures. A genetic linkage map of the population was created with JoinMap 3.0 using genotypic data from 532 amplified fragment length polymorphism (AFLP) and 39 SSR markers across the 137 lines. AFLP markers clustered in regions heterogeneous between the parents, and 442 AFLP markers were placed onto the 12 chromosome pairs of rice using SSR markers of known location as anchors. The remaining 90 AFLP markers were linked into 13 relatively small groups for which the genomic location remains unknown at this time. Using simple regression, the proportion of brokens in the brown rice after de-hulling and before milling (PB = Pre-broken), explained 59.2% of the variance (R2) in WR/TR, indicating this is an excellent predictor of milling yield. Chalkiness in brown rice explained 19.2% of the variance in PB while apparent amylose explained 9.6% of the variance. BR explained 67.9% of the variance in TR but was unrelated to WR and WR/TR. TR explained 14.1% of the variance in WR. Chalk and apparent amylose explained 20.4% and 14.8% of the variance in WR/TR, respectively. Moreover, length, width and thickness accounted for only for 0.3%, 2.6%, and 0.5% of the variance in WR/TR, respectively; indicating that grain shape in this long grain cross had little impact on milling yield. Significant QTLs were identified for BR, TR, TR/BR, WR and WR/TR, with qBR-3, qTR-2, qTR-3, qTR/BR-2, qWR-6 and qWR/TR-6 explaining 23.5%, 14.6%, 12.2%, 13.8%, 13.7% and 14.7% of their phenotypic variances, respectively. Multiple significant QTLs for length, width, thickness, and chalkiness were also identified. The percentage of phenotypic variance explained by the single largest contributing QTL for these traits was 22.4%, 16.3%, 21.2%, 12.6%, respectively. The Waxy locus (qWR/TR-6: RM190) was directly associated with 14.7% of the variance in WR/TR. RM190 was also the largest contributing locus to the variance in PB, explaining up to 15.4%. As a next step, we will perform QTL mapping within RM190 genotypic classes to identify additional loci with significant effects on WR/TR and PB. In conclusion, we have determined that chalkiness and, surprisingly, amylose content had a significant impact on milling yield in this long grain cross. We are converting the AFLP markers associated with chalkiness in this population into micro-satellite (SSR) markers for testing and verification in U.S. rice breeding programs. Although grain shape components did not have a major impact on milling in this cross, we also identified significant QTLs for grain length, width, and thickness that may be useful in breeding programs. Moreover, we determined that PB is a simple and efficient means of evaluating progeny for milling yield potential. We are continuing our investigation to identify additional markers for milling yield in Southern U.S. long grain rice.