Intensive rice breeding over the past 50 y has dramatically increased productivity especially in the indica subspecies, but our knowledge of the genomic changes associated with such improvement has been limited. In this study, we analyzed low-coverage sequencing data of 1,479 rice accessions from 73 countries, including landraces and modern cultivars. We identified two major subpopulations, indica I (IndI) and indica II (IndII), in the indica subspecies, which corresponded to the two putative heterotic groups resulting from independent breeding efforts. We detected 200 regions spanning 7.8% of the rice genome that had been differentially selected between IndI and IndII, and thus referred to as breeding signatures. These regions included large numbers of known functional genes and loci associated with important agronomic traits revealed by genome-wide association studies. Grain yield was positively correlated with the number of breeding signatures in a variety, suggesting that the number of breeding signatures in a line may be useful for predicting agronomic potential and the selected loci may provide targets for rice improvement.) is one of the most important cereal crops in the world. There have been landmark achievements in rice improvement over the past 50 y, especially in the indica subspecies. A major breakthrough resulted from the independent development of a series of semidwarf varieties in China and by the International Rice Research Institute (IRRI) in the 1950s and 1960s, leading to the "green revolution" in rice. Since then, semidwarfness has been a basic characteristic for almost all modern varieties. Based on semidwarf varieties, improvement for other traits, such as abiotic stress resistance, broad-spectrum resistances to biotic stresses, and better grain quality, has also been achieved. Another major breakthrough stemmed from the exploitation of hybrid vigor in China (1), resulting in the largescale adoption of hybrid rice since the 1970s. Jointly, these breakthroughs have greatly increased rice productivity in the past several decades globally.Genomic studies in recent years have identified a large number of loci that were under selection during rice domestication (2). However, there has been very limited study to identify loci or genomic regions that have been under selection due to breeding. Next-generation sequencing technologies have enabled sequencing of a large number of rice accessions at relatively low cost, providing opportunities to inspect the genomic regions selected in the history of crop improvement. Meanwhile, genome-wide association studies (GWAS) have provided an effective approach to analyze the genetic architecture of complex traits and allow identification of candidate genes for further improvement of agronomically important traits (3,4).In this study, we analyzed low-coverage sequencing data of 1,479 rice accessions, which revealed a large number of differentially selected regions associated with breeding efforts between two major subpopulations in indica. These selected regi...