*3.3. Population Structure and Linkage Disequilibrium*

The diversity panel could be divided into two subgroups, the *Japonica* subpopulation and the other derived from the *Indica* accessions, whose characterization were largely consistent with geographic origins. Besides, admixture between *Indica* and *Japonica* were also observed in the present study (Figure 3c). Numerous studies have shown that the lack of appropriate correction for population structure can lead to spurious maker-trait associations (MTAs). Consequently, to eliminate spurious MTAs resulting from population structure, an MLM implemented in GAPIT were adopted for association analysis in the current study. PCA analysis indicated that the top three PCs could explain 28.5%, 8.2% and 3.2% of the total variation of population structure, respectively, and this panel consists of two subgroups (Figure 3a). The neighbor-joining (NJ) tree showing two clades also represented the two subpopulations (Figure 3b).

The decay of LD along physical distances was computed for the 208 rice accessions. A scatter *r*2 against physical distance showed a clean pattern of LD decay in the 208 rice accessions. A critical value of the determination coefficients *r*2 > 0.2 was determined to be the threshold for LD decay and the LD decay distance was about 150 Kb (Figure 3d).

**Figure 3.** Population structure and LD decay of 208 rice accessions. (**a**) Principal component plot, different colors represent different ingredients; (**b**) neighbor-Jointing tree; (**c**) structure; (**d**) LD decay analysis.
