Pyramiding effects of strong-culm genes on traits associated with lodging resistance
Figure 1 shows the breeding process for the pyramiding lines. Pyramiding lines with double NILs (SCM1 + 2, SCM1 + 3, SCM1 + 4, SCM2 + 4, SCM2 + 3, SCM3 + 4), triple NILs (SCM1 + 2 + 4, SCM1 + 2 + 3, SCM1 + 3 + 4, SCM2 + 3 + 4), and a quadruple NIL (SCM1 + 2 + 3 + 4) were developed. To investigate the effect of QTL pyramiding on strong-culm related traits, we compared the number of QTL pyramiding for each trait related to bending-type lodging resistance using single NILs, pyramiding lines and Koshihikari (Fig. 2). The mean Flexural rigidity (FR) was higher in double, triple, and quadruple pyramiding lines, compared to single NILs and Koshihikari, and increased proportionally to the number of QTL pyramiding (Fig. 2c). For the component traits of FR, secondary moment of inertia (SMI) increased with QTL pyramiding number, while the Young’s modulus (YM) decreased (Fig. 2a, b).
Figure 2d-f shows the M and its component traits, SM and BS, which are indicators of breaking-type lodging resistance. The mean M in the triple and quadruple pyramiding lines was significantly higher than in Koshihikari. For the component traits, the cross-section modulus increased with the QTL pyramiding number, while the BS decreased.
For each of these traits, we examined which combination of QTLs had the highest pyramiding effect on the lodging resistance-related traits (Fig. 3). Figure 3A-C shows the bending-type lodging resistance-related traits for each combination of QTLs. In the double pyramiding lines, the SMI of SCM1 + 2, SCM1 + 3, and SCM2 + 4 was significantly increased compared with Koshihikari, most notably in SCM1 + 2 and SCM1 + 3. In the triple pyramiding lines, the pyramiding effect was significantly higher in all three lines than in Koshihikari, especially in SCM1 + 2 + 3 and SCM1 + 2 + 4 (Fig. 3c). The Young’s modulus of the almost all pyramiding lines was significantly reduced compared with Koshihikari, with only SCM3 + 4 not being significantly different. Among the double pyramiding lines, SCM2 + 4 had the lowest Young’s modulus. In the triple pyramiding lines, Young’s modulus was highest in SCM1 + 3 + 4 and lowest in SCM1 + 2 + 4 (Fig. 3b).
Figure 3D-F shows the breaking-type lodging resistance-related traits for each combination of QTLs. The SM of SCM1 + 2, SCM1 + 3, SCM2 + 3, and SCM2 + 4 was significantly increased compared with Koshihikari, and the pyramiding effect of SCM1 + 2 and SCM1 + 3 was particularly high in the double pyramiding lines. In the triple pyramiding lines, SM was significantly increased in all lines, and the pyramiding effect was particularly high in SCM1 + 2 + 3 and SCM1 + 3 + 4 compared with Koshihikari. Bending stress was significantly reduced in almost all lines compared to Koshihikari, with only SCM1 + 3 not being significantly different. Among the double pyramiding lines, SCM2 + 4 had the lowest BS. In the triple pyramiding lines, BS showed the same trend as observed with Young’s modulus. Among the double pyramiding lines, the M of SCM1 + 2 and SCM1 + 3 was significantly increased compared with Koshihikari. Among the triple pyramiding lines, M was significantly increased compared with Koshihikari in all lines except SCM2 + 3 + 4.
Figure 4 shows a photograph of representative main culms in the quadruple pyramiding line SCM1 + 2 + 3 + 4 and Koshihikari. It can be seen that SCM1 + 2 + 3 + 4 had a thick culm phenotype compared with Koshihikari.
Additional effects and epistatic interactions for strong-culm genes
Multiple linear regression analysis was then used to examine the contribution of each QTL for strong-culm traits and epistasis among QTLs in single NILs and pyramiding lines (Supplementary Table 1).
Table 1 shows the estimated interactions between QTLs in SMI and SM. The estimated SMI increased for SCM1, SCM2, SCM3 and SCM4 in this order, indicating that the two QTLs derived from Habataki had a larger incremental effect on SMI. Moreover, SCM1, SCM2, SCM3 and SCM4 had increasing contributions to SM in that order, similar to the results observed for SMI.
Supplementary Fig. 1 shows the relationship between the estimated and measured SMI and SM excluding interactions for Koshihikari and the pyramiding lines. For SMI, the regression equation was y = 0.922 x + 4.564 with a coefficient of determination of 0.850. For SM, the regression equation was y = 0.995 x + 0.684 with a coefficient of determination of 0.873. The close proximity of y = x for both thick culm traits, and the high coefficients of determination indicated that estimates of the individual QTLs, excluding interactions among QTLs, explained the majority of the thick culm traits in the pyramiding lines. These results indicated that culm thickness was determined by a simple additive effect, and did not involve epistasis.
Evaluation of the pyramiding effects of strong-culm genes on lodging resistance
To evaluate the effect of the pyramiding of strong-culm QTLs on lodging resistance, we conducted an artificial typhoon test and measured the degree of lodging (Supplementary Fig. 2). The artificial typhoon test was performed on Koshihikari and three pyramiding lines (SCM1 + 2 + 3, SCM1 + 3 + 4, and SCM1 + 2 + 3 + 4), which were more effective in QTL pyramiding on strong-culm associated traits. Compared with Koshihikari, only a small degree of curvature in the pyramiding lines was observed (Supplementary Fig. 2a). Thus, the angle of curvature of the pyramiding lines was investigated. The results are shown in Supplementary Fig. 2b. Among the four lines tested, Koshihikari had the largest bending angle (43°) after the artificial typhoon test. In contrast, the three pyramiding lines had a smaller angle of bending than Koshihikari. In particular, the bending of SCM1 + 3 + 4 was significantly reduced by 58% compared to Koshihikari.
To evaluate the occurrence of lodging in actual paddy fields, we measured the degree of lodging (Fig. 5). Figure 5a shows representative photographs of Koshihikari and SCM1 + 2 + 3 + 4 at the late maturing stage. In Koshihikari, lodging occurred after typhoon exposure, whereas no lodging occurred in SCM1 + 2 + 3 + 4. The degree of lodging was greater in Koshihikari due to the high degree of plant bending; however, no lodging occurred in the other lines, except for SCM2 (Fig. 5b).
Causal factors of strong culm in pyramiding lines
To investigate the causal factors of high FR and M in the pyramiding lines, we focused on the structural properties in cortical fiber tissue (Fig. 6). The correlation coefficient between FR and the thickness of the cortical fiber tissue was 0.82, and the correlation coefficient between M and the thickness of the cortical fiber tissue was 0.80. In both cases, a strong positive correlation was observed (Fig. 6a, b). The lines containing SCM1 exhibited thicker cortical fiber tissue, large FR and M. The lines with a significantly thicker cortical fiber tissue compared to Koshihikari were SCM1, SCM1 + 2, SCM1 + 3, SCM1 + 2 + 3, SCM1 + 3 + 4, and SCM1 + 2 + 3 + 4, all of which contained SCM1 (Fig. 6c). Figure 6d shows the number and density of cell layers in the cortical fiber tissue of lines containing SCM1. The number of cell layers was significantly increased in SCM1, SCM1 + 2, SCM1 + 3, and SCM1 + 2 + 3 + 4, and the cell density was significantly increased in SCM1, SCM1 + 2 + 3, and SCM1 + 3 + 4 compared to Koshihikari.
Pleiotropic pyramiding effects of strong-culm genes on traits associated with yield components
To investigate the effects of strong-culm QTL pyramiding on the pleiotropic expression of yield component traits, we compared the yield and yield components among Koshihikari and pyramiding lines with a superior lodging resistance (Supplementary Table 2). In the single NILs (SCM1, SCM3, and SCM4), the panicle number tended to decrease compared to Koshihikari in both years, especially in SCM1 and SCM3 in 2018, with 19% and 17% decreases in panicle number, respectively. In the pyramiding lines, the panicle number of SCM1 + 2 + 3, SCM1 + 3 + 4, and SCM1 + 2 + 3 + 4 was lower than that of Koshihikari. In SCM1 + 2 + 3 and SCM1 + 2 + 3 + 4, panicle number decreased significantly. In particular, the panicle number in SCM1 + 2 + 3 + 4 decreased significantly in both years, 31% in 2017 and 44% in 2018, compared with that in Koshihikari. On the other hand, the grain number per panicle increased in the four single NILs. Notably, the grain number of SCM1 significantly increased by 36% in 2017 and 35% in 2018 compared with that of Koshihikari. The grain number per panicle of the pyramiding lines was significantly higher than that of Koshihikari in both years, and this increase was greater than those of the single NILs. In particular, the grain number per panicle in SCM1 + 2 + 3 + 4 increased by 61% in 2017 and 68% in 2018 compared to Koshihikari. Although large differences in the yield components were found among Koshihikari and pyramiding lines, there were no differences in grain yield between Koshihikari and the pyramiding lines.
Finally, we investigated the eating quality in the pyramiding line SCM1 + 3 + 4 (Supplementary Table 3). SCM1 + 3 + 4 had a premium eating quality with a low protein, which is similar to Koshihikari. Strong culm genes of pyramiding lines did not affect to their eating quality in the Koshihikari genetic background. We developed the new variety ‘Sakura prince’ with a strong culm derived from SCM 1 + 3 + 4 using marker assisted selection.