Growth-defence trade-off in rice: fast-growing and acquisitive genotypes have lower expression of genes involved in immunity

Abstract

Plant ecologists and molecular biologists have long considered the hypothesis of a trade-off between plant growth and defence separately. In particular, how genes thought to control the growth-defence trade-off at the molecular level relate to trait-based frameworks in functional ecology, such as the slow-fast plant economics spectrum, is unknown. We grew 49 phenotypically diverse rice genotypes in pots under optimal conditions and measured growth-related functional traits and the constitutive expression of 11 genes involved in plant defence. We also quantified the concentration of silicon (Si) in leaves to estimate silica-based defences. Rice genotypes were aligned along a slow-fast continuum, with slow-growing, late-flowering genotypes versus fast-growing, early-flowering genotypes. Leaf dry matter content and leaf Si concentrations were not aligned with this axis and negatively correlated with each other. Live-fast genotypes exhibited greater expression of OsNPR1, a regulator of the salicylic acid pathway that promotes plant defence while suppressing plant growth. These genotypes also exhibited greater expression of SPL7 and GH3.2, which are also involved in both stress resistance and growth. Our results do not support the hypothesis of a growth-defence trade-off when leaf Si and leaf dry matter content are considered, but they do when hormonal pathway genes are considered. We demonstrate the benefits of combining ecological and molecular approaches to elucidate the growth-defence trade-off, opening new avenues for plant breeding and crop science.

Keywords: Defence gene; growth–defence trade-off; intraspecific variation; plant defence; plant economics spectrum; plant functional trait; plant immunity; rice (Oryza sativa); silica; silicon.

Fig. 1.

Principal component analyses (PCAs) based on correlation matrices of the main functional traits (A) and gene expression (B) for 49 rice genotypes. The relative growth rate (RGR) and age at flowering are supplementary quantitative variables for visualization (grey, dashed lines), and have no influence on the PCA results. The colour gradient indicates regions of highest (red) to lowest (white) occurrence of probability of genotypes in the traits/genes space, and contour lines show 0.50, 0.90, and 0.95 quantiles (see Díaz et al., 2016 for the method). The results of the PCAs are presented in Supplementary Tables S3, S6. Both PCAs were run on log-transformed data. LDMC, leaf dry matter content; Y(II), chlorophyll fluorescence.

Fig. 2.

Relationship between leaf silicon concentrations and leaf dry matter content. (A) Standardized major axis regression line and statistics of the bivariate relationship between leaf silicon (Si) concentrations and leaf dry matter content (LMDC) among 49 rice genotypes. (B) To avoid spurious correlations, we also examined the relationship between leaf [Si] and silica-free LDMC. Silica-free LDMC was calculated as follows: (leaf dry weight−silica weight)/(leaf fresh weight−silica weight). The weight of silica in the leaves used for LDMC was calculated using leaf [Si]. Si was converted into silica by multiplying by 2.14 and assuming a 10% mean water content and a 5% content of other elements (Blecker et al., 2006). Both relationships were very similar, with a slightly greater explanatory power for the relationship between leaf [Si] and corrected LDMC.

Fig. 3.

Relationships between PCA (shown in Fig. 1) axis scores. (A) Relationships between genotype scores on PC1 of the PCA considering functional traits (slow–fast continuum) and both PCs of the PCA considering gene expression. (B) Relationships between genotype scores on PC2 of the PCA considering functional traits (Si–LDMC axis) and both PCs of the PCA considering gene expression. Standardized major axis regression lines and statistics of bivariate relationships are given.

Fig. 4.

Relationships between the expression of NPR1, GH3.2, and SPL7 and the genotype scores on PC1 of the PCA considering functional traits (slow–fast continuum) (A) and the genotypes relative growth rates (RGR) (B). Standardized major axis regression lines and statistics of bivariate relationships are given. y-axes are on a logarithmic scale.