AbstractIt has been estimated that up to 90% of human exposure to cadmium is through food, and that cadmium within rice grains can be a major contributor to that dietary source. In this study genome wide association mapping was conducted on the Bengal and Assam Aus Panel (BAAP) of rice to identify quantitative trait loci and candidate genes for lowering grain cadmium. Field experiments were conducted over two years under two different irrigation systems: continually flooded and alternate wetting and drying (AWD). There was significant effects of water treatment, genotype, and genotype by water treatment interaction. Importantly, AWD increased grain cadmium, on average, by 49.6% and 108.8% in year 1 and 2 respectively. There was between 4.6 and 28 fold variation in cadmium concentration. A total of 58 QTLs were detected but no loci are clearly specific to one water regime despite approximately 20% of variation attributable to genotype by water regime interaction. A number of QTLs were consistent across most water treatments and years. These included QTLs on chromosome 7 (7.23–7.61, 8.93–9.04, and 29.12–29.14 Mbp), chromosome 5 (8.66–8.72 Mbp), and chromosome 9 (11.46–11.64 Mbp). Further analysis of the loci on chromosome 7 (8.93–9.04 Mbp), identified the candidate gene OsNRAMP1, where cultivars with a deletion upstream of the gene had higher concentrations of cadmium compared to the cultivars that did not have the deletion. The distribution of alleles within the BAAP suggest this QTL is easily detected in this population because it is composed of aus cultivars. Local genome cluster analysis suggest high Cd alleles are uncommon, but should be avoided in breeding.