*2.3. QTL Mapping and Heritability Estimation*

Calculation results of the average allele frequency (AF) value for each marker showed that most of the identified SNP and short InDel markers had an expected AF value of around 0.5 in the four sub-pools ER-1, ER-2, ES-1, and ES-2 (Supplementary Figure S2), indicating no severe segregation distortion of the markers as a whole. Allele frequency difference (AFD) value between ER (ER-1 + ER-2) and ES (ES-1 + ES-2) pools was calculated, and four positive AFD peaks were detected in the fitted curve and exceeded the threshold (0.165 at the overall significance level of *p* < 0.05) (Figure 2A). Subsequently, unpaired t-tests were performed for the two replicated sub-pools for ER and ES. The *p*-values of each marker were estimated, and the peaks of negative logarithmic *p* value (NLP) (Figure 2B) were consistent with the peaks in AFD curve. These results suggest that there were four candidate QTLs located in these regions. The confidence intervals of the four QTLs located on chromosome 4, 7, 8, and 11 were estimated (Figures 2A and 3, Table 2), and the four QTLs were named *qBBR-4*, *qBBR-7*, *qBBR-8*, and *qBBR-11*, respectively.

Heritability estimation showed that the additive heritability and dominance heritability of each QTL varied by 0.89%−2.09% and 0.01%−7.82%, respectively (Table 2). Among the four QTLs, *qBBR-4* had the largest effect, with the biggest additive heritability at about 2.09% (Table 2), suggesting that *qBBR-4* is a major QTL involved in basal resistance to blast disease.

**Figure 2.** BSA-seq-based identification of four candidate quantitative trait loci (QTLs) conferring basal resistance to rice blast disease. (**A**) Allele frequency difference (AFD) graph from BSA-seq analysis. The horizontal orange dashed lines indicate the threshold (±0.165) at the overall significance level of *p* < 0.05. QTL positions estimated are indicated by filled triangles. AFD was obtained by subtraction of allele frequency (AF) of the extremely susceptible (ES) pool from that of the extremely resistant (ER) pool. (**B**) *T*-test verification for the two replicated sub-pools for the ER pool and the ES pool. The average negative logarithmic *p*-values of the markers were smoothed by sliding window (size = 3000 kb and step = 10 kb) across each chromosome. The peaks of the *p*-value graph were consistent with those in the AFD graph.

**Figure 3.** Estimates of confidence intervals of the four QTLs, *qBBR-4* (**A**), *qBBR-7* (**B**), *qBBR-8* (**C**), and *qBBR-11* (**D**). The horizontal orange dashed lines indicate the threshold (0.165) at the overall significance level of *p* < 0.05. The light green areas indicate 95% confidence intervals of the QTLs. Previously reported *R* genes or QTLs within or closely near the confidence intervals of *qBBR-4*, *qBBR-8*, and *qBBR-11* are indicated on the right. AFD: allele frequency difference.

**Table 2.** Estimates of position and heritability of the four identified QTLs.


<sup>a</sup> Maximum value of the peak of the AFD curve; <sup>b</sup> Chromosome position of the peak of the AFD curve; <sup>c</sup> Estimated based on the 95% confidence; <sup>d</sup> The most significant *p*-value of the peak, which was converted to negative logarithmic *p* (NLP) value; <sup>e</sup> Heritability attributed to additive effect of the QTL; <sup>f</sup> Heritability attributed to dominance effect of the QTL.

#### *2.4. Identification of a New Haplotype of pi21 as a Candidate Gene of qBBR-4*

To further refine candidate genes involved in basal resistance to blast disease, we searched previously reported *R* genes or QTLs within the confidence intervals of the four QTLs (Figure 3). While no previously reported *R* genes or QTLs were identified within or near the confidence interval of *qBBR-7* (Figure 3B), a previously identified *Pia* [26] gene was located near the confidence interval of *qBBR-11* (Figure 3D), and there were several *R* genes or QTL, including *Pi-11(t)* [27], *Pi-29(t)* [28], *Pi33* [29], *Pi-GD-1(t)* [30], and *qDSF8* [3], in the *qBBR-8* region (Figure 3C). Interestingly, *qBBR-4* was observed to be co-localized with a cloned recessive durable blast disease resistance QTL *pi21* (Figure 3A). *Pi21* has been found to have at least 12 variants (haplotypes A to L) based on InDel polymorphisms in the proline-rich region. While haplotype L containing double deletions of 21 bp and 48 bp (*pi21(-21*/*-48)*) resulting in deletions of the core motif "PxxPxxP" in the proline-rich region (Figure 4A) was identified, conferring durable blast disease resistance due to loss of function of *Pi21*, the other haplotypes carrying one of the two deletions or two smaller deletions did not confer high basal resistance to blast disease [16].

Sequence analysis of the *Pi21* alleles in LXG and 02428 showed that while the allele in LXG (*Pi21-LXG*) had a 27 bp deletion, the *pi21-2428* allele had double deletions of 30 bp and 33 bp, resulting in deletions of 10 aa and 11 aa of the core "PxxPxxP" motif, the same as in the *pi21(-21*/*-48)* allele (Figure 4A,B). Both *Pi21-LXG* and *pi21-2428* were different from the 12 identified haplotypes. Genotypic analysis of 61 individual ER and 59 individual ES plants showed that all *pi21-2428* homozygous seedlings showed resistance to blast disease. In contrast, about 79% (41 out of 52) of the *Pi21-LXG* homozygous, and 39% (18 out of 46) of *Pi21-LXG*/*pi21-2428* heterozygous seedlings showed susceptible to blast disease (Figure 4C). These results support *pi21-2428* as a candidate gene of *qBBR-4*, and suggest that while the 27 bp deletion did not affect *Pi21-LXG* function, double deletions of the 30 bp and 33 bp sequences could cause a defect in *pi21-2428* function, leading to high basal resistance to blast disease.
