QTL mapping for agronomic traits using multi-parent advanced generation inter-cross (MAGIC) populations derived from diverse elite indica rice lines

Abstract

In rice many QTL have been reported for yield and related traits using various bi-parental mapping populations. However, few of the well characterized genes/QTL have been successfully used in breeding for improved trait performance. One of the reasons is that the mapping populations used are irrelevant to breeding. Association mapping has also been used in identifying marker-trait associations that are effective in more complex background. However, the results from association mapping using diversity panels are difficult to be exploited in breeding, since most of the accessions had poor performance in many important agronomic traits. In this study, a genome-wide association study (GWAS) was performed using three multi-parent advanced generation inter-cross (MAGIC) populations derived from elite indica lines (DC1, DC2 and 8-way) to identify QTL for 14 traits including yield, yield components and other related traits. The three MAGIC populations were phenotyped in the dry season (DS) and wet season (WS) of 2014 at the headquarters of the International Rice Research Institute (IRRI) and genotyped with a Rice SNP chip containing 4,500 markers. A total of 26 QTL on all chromosomes except 7, 9 and 11 were identified for 10 traits in the DS or WS. Six, two, 12, 10 and 20 out of the 26 QTL were identified in the DC1, DC2, 8-way, DC12 (DC1+DC2) and RMPRIL (DC1+DC2+8-way) populations, respectively. Nine of the QTL corresponded to known QTL/genes, including qFLW4 for FLW, qTGW3 and qTGW5 for TGW, qFGN4 for FGN, qSBN4 for SBN, qPH1.1 and qPH1.2 for PH, qHD3 and qHD6 for HD. All these nine QTL were identifiable using the RMPRIL population while only eight, seven, six and one could be identified using the 8-way, DC12, DC1 and DC2 populations, respectively. The 8-way population was more powerful than the DC1, DC2 and DC12 populations. The joint analysis, which combines different populations (e.g., DC12 and RMPRIL), increased the number of QTL identified and mapping resolution. The study showed that MAGIC populations derived from diverse elite parental lines can be used to detect QTL, which are ideal for linking gene identification and practical breeding. GWAS using such MAGIC populations had higher detection power (compared to assembled populations) and higher resolution (compared to biparental populations). The identified QTL are directly applicable, since the populations are good breeding populations as well.