Genotype-by-environment interaction for SOC
We conducted multi-environment trials with a diversity panel of 505B. napus lines to examine the effect of genotype-by-environment interaction (G×E) on SOC. We carried out these trials at three different sites (Chengdu, CD; Wuhan, WH, and Hefei, HF), spanning a wide geographical range with diverse temperature and precipitation patterns (Fig. 1a, b; Supplementary Tables S1, S2). To account for potential year-to-year variations in temperature and precipitation, we conducted these trials for up to 4 years at each site. We assessed the contributions of genotype and environment to the observed variation in SOC in the multi-environment trials. SOC values among all B. napus lines show Pearson’s correlation coefficients from 0.27 to 0.68 across eight environments, indicating the influence of the environment on SOC (Fig. S1a, b). Looking at the influence of genotype, we obtained correlation coefficient values from −0.87 to 0.99 for all pairs of inbred lines (Fig. S1c). An analysis of variance for SOC across all eight environments indicates that environment, genotype, and G×E have similar contributions, explaining 23.75%, 25.26%, and 18.90% of the SOC variation, respectively (Supplementary Tables S5, S6). Thus, genotype and environment both influence SOC in this population.
We then analyzed the phenotypic variation in SOC using a joint regression analysis. For each genotype, we captured SOC across eight environments by a linear regression model with two parameters: the intercept, which represents the average response of a given genotype to the eight environments, and the slope, which represents the plasticity of that genotype to different environments (Fig. 1c). The resulting models show a strong correspondence with the SOC observations, as indicated by the goodness-of-fit (R 2 = 0.56, on average). We observed substantial variation in both the intercept (ranging from 39.44% to 46.62% oil) and the slope (ranging from -0.59 to 2.36) among different genotypes. To investigate whether these variations are influenced by population structure, we divided ourB. napus lines into five subgroups based on genome-wide single nucleotide polymorphisms (SNPs) and examined SOC variation within each subgroup. We observed no effect from population structure on intercept or slope, with little correlation between these two parameters in either the overall population (r = 0.01) or individual subgroups (r = 0.01–0.24, Fig. S2). This finding suggests that genotypes with higher or lower average responses to all eight environments do not necessarily exhibit higher or lower plasticity to different environments.
We further investigated crossover occurrence by analyzing the proportion of pairs of lines experiencing changes in ranking, a prominent type of G×E(Xiong et al., 2021). Crossover occurrence is not evenly distributed across environmental means, with the highest occurrence observed near the center of the distribution (Fig. 1d). To examine whether this distribution is associated with variable genetic architecture across environments, we assessed the SNP-based heritability, which quantifies the proportion of phenotypic variation attributed to genetic variation. SNP-based heritability varies among environments, ranging from 0.13 to 0.56, and tends to be lower in environments closer to the center of the environmental mean (Supplementary Table S7). These estimations reveal that the variable genetic architectures across environments explain the G×E effects observed in this B. napus population.