Fig. 3: Quantitative complementation test of the RDP1 gene.

Violin plots with means and standard errors of means indicated by red bold bars and boxes, respectively. a, b Pollen number differences between wild-type and homozygous plants of a frameshift allele generated by the CRISPR/Cas9 technique in the Bor-4 background (a Numbers of flowers pollen-counted: RDP1^Bor/RDP1^Bor, N = 89; rdp1^Bor/rdp1^Bor, N = 77) and in the Uod-1 background (b Numbers of flowers pollen-counted: RDP1^Uod/RDP1^Uod, N = 47; rdp1^Uod/rdp1^Uod, N = 43) (same data sets with Fig. 2h). c The difference in the effect on pollen number by two natural alleles, RDP1^Bor and RDP1^Uod. Pollen number of plants with RDP1^Uod was significantly lower than that of plants with RDP1^Bor (nested analysis of variance; P = 4.85 × 10^–8; Numbers of flowers pollen-counted: RDP1^Bor/rdp1^Uod, N = 468 from 26 individuals; rdp1^Bor/RDP1^Uod, N = 368 from 20 individuals). The two alleles were compared in the heterozygous state with a frameshift CRISPR/Cas9 allele, with otherwise identical genetic backgrounds. F₁ plants were obtained from the cross of two heterozygotes, RDP1^Bor/rdp1^Bor and RDP1^Uod/rdp1^Uod. Source data are provided as a Source Data file.