Since domestication, rice has cultivated in a wide range of latitudes with different day lengths. Selection of diverse natural variations in heading date and photoperiod sensitivity is critical for adaptation of rice to different geographical environments. To unravel the genetic architecture underlying natural variation of rice flowering time, we conducted a genome wide association study (GWAS) using several association analysis strategies with a diverse worldwide collection of 529 O. sativa accessions. Heading date was investigated in three environments under long-day or short-day conditions, and photosensitivity was evaluated. By dividing the whole association panel into subpopulations and performing GWAS with both linear mixed models and multi-locus mixed-models, we revealed hundreds of significant loci harboring novel candidate genes as well as most of the known flowering time genes. In total, 127 hotspots were detected in at least two GWAS. Universal genetic heterogeneity was found across subpopulations. We further detected abundant interactions between GWAS loci, especially in indica. Functional gene families were revealed from enrichment analysis of the 127 hotspots. The results demonstrated a rich of genetic interactions in rice flowering time genes and such epistatic interactions contributed to the large portions of missing heritability in GWAS. It suggests the increased complexity of genetic heterogeneity might discount the power of increasing the sample sizes in GWAS.

自驯化以来，稻米已在不同日长的广泛纬度下栽培。选择抽穗期和光周期敏感性的各种自然变化对于水稻适应不同地理环境至关重要。为了揭示水稻开花时间自然变异的遗传结构，我们使用了多种关联分析策略，对全球529 O. sativa进行了收集，进行了全基因组关联研究（GWAS）。加入。在长日或短日条件下的三种环境中研究了抽穗期，并评估了光敏性。通过将整个关联面板划分为亚群，并使用线性混合模型和多位点混合模型执行GWAS，我们揭示了数百个重要的基因座，这些基因座包含新的候选基因以及大多数已知的开花时间基因。总共在至少两个GWAS中检测到127个热点。跨亚群发现普遍的遗传异质性。我们进一步检测到了GWAS基因座之间的大量相互作用，尤其是在印度。通过对127个热点的富集分析揭示了功能基因家族。结果表明水稻开花时间基因中存在丰富的遗传相互作用，而这种上位性相互作用导致了GWAS中大部分遗传力缺失。这表明遗传异质性增加的复杂性可能抵消了GWAS中增加样本量的力量。