Natural variation in Arabidopsis shoot branching plasticity in response to nitrate supply affects fitness

Abstract

The capacity of organisms to tune their development in response to environmental cues is pervasive in nature. This phenotypic plasticity is particularly striking in plants, enabled by their modular and continuous development. A good example is the activation of lateral shoot branches in Arabidopsis, which develop from axillary meristems at the base of leaves. The activity and elongation of lateral shoots depends on the integration of many signals both external (e.g. light, nutrient supply) and internal (e.g. the phytohormones auxin, strigolactone and cytokinin). Here, we characterise natural variation in plasticity of shoot branching in response to nitrate supply using two diverse panels of Arabidopsis lines. We find extensive variation in nitrate sensitivity across these lines, suggesting a genetic basis for variation in branching plasticity. High plasticity is associated with extreme branching phenotypes such that lines with the most branches on high nitrate have the fewest under nitrate deficient conditions. Conversely, low plasticity is associated with a constitutively moderate level of branching. Furthermore, variation in plasticity is associated with alternative life histories with the low plasticity lines flowering significantly earlier than high plasticity lines. In Arabidopsis, branching is highly correlated with fruit yield, and thus low plasticity lines produce more fruit than high plasticity lines under nitrate deficient conditions, whereas highly plastic lines produce more fruit under high nitrate conditions. Low and high plasticity, associated with early and late flowering respectively, can therefore be interpreted alternative escape vs mitigate strategies to low N environments. The genetic architecture of these traits appears to be highly complex, with only a small proportion of the estimated genetic variance detected in association mapping.

Fig 1. Genetic and non-genetic components of trait variation.

Each trait’s variance was decomposed into genetic and non-genetic components using linear mixed models: genotype-specific effects both on high (HN) and low (LN) nitrate; nitrate-specific effects (environmental component); the response of each genotype to nitrate supply (genotype-by-environment interaction); and unexplained variance (residual). (A) shows the proportion of variance attributed to each component for each trait and (B) shows the magnitude of this variance relative to each trait’s mean, using the coefficient of variation (the estimated variance divided by the squared mean of the respective trait). Measurements of height and branch number were taken when the plants had two expanded siliques. For the accessions, the total number of branches was also scored at the senescence stage (sen), together with the total number of siliques. Variance components were estimated using data from 374 MAGIC lines and 297 (2 silique stage) or 278 (senescence stage) accessions with n = 4–8 replicates for each line on each nitrate treatment (median n = 8). Detailed results from these models are shown in S2 Table.

Fig 2. Relationship of trait means between the two nitrate treatments.

Relationship between the mean trait value of each genotype grown on high (HN) or low (LN) nitrate supply shown as a scatter plot (left), as a reaction norm plot (middle) and as a histogram of relative plasticity (right). The latter shows the “relative distance plasticity index” of [105], which varies between -1 and 1 and with zero indicating no plasticity. Genotypes plotted in red on the left panel show significant plastic responses to N, as assessed by a Wilcoxon rank-sum test (false discovery rate < 5%). There were none for flowering time. The reaction norm plots are coloured by the relative plasticity index of each genotype, with the corresponding colour scale shown along the x-axis of the respective histograms. (A, B) Data from 374 MAGIC lines. (C, D) Data from 297 natural accessions. All plants were scored at the 2-silique stage, with means from n = 4–8 replicates per line in each nitrate treatment (median n = 8). Pearson’s correlation coefficient (r) is shown in each panel with the 95% confidence interval shown in brackets. The dashed line on the left panels is the identity line (x = y). Note the log-scale on the flowering time plots.

Fig 3. Trait correlations in the MAGIC lines.

(A) Correlation between mean shoot branching plasticity and mean number of branches under low (LN) and high (HN) N supply. Pearson’s correlation coefficient (r) for each treatment is shown in the panels. The 95% confidence intervals are: r(LN) = -0.57 [-0.63, -0.49], p ~ 1⁻³³; r(HN) = 0.65 [0.59, 0.71], p ~ 10⁻⁴⁶. (B) Reaction norm plots of the number of branches for the 25 least and most plastic lines. Lines are coloured by the average days to flowering of each MAGIC line. (C) Correlation between mean days to flowering and branching plasticity. The blue line is a smoothed trend fitted by local regression (LOESS). The dotted line shows a cut-off of 25 days to flowering on LN. Pearson’s correlation coefficients are shown for all 374 lines (r = 0.068 [-0.033, 0.17], p = 0.19) and for 258 early-flowering lines only (r = 0.67 [0.60, 0.73], p ~ 1⁻³⁵). In all panels, data are means from n = 4–8 replicates of each line (median n = 8). Note the log-scaled x axis on panel C.

Fig 4. Trait correlations in the natural accessions.

(A) Correlation between mean shoot branching plasticity and mean number of branches under low (LN) and high (HN) N supply at the 2-silique stage for 297 accessions. Pearson’s correlation coefficient (r) is shown for each nitrate treatment. The 95% confidence intervals are: r(LN) = -0.39 [-0.48, -0.29], p ~ 10⁻¹²; r(HN) = 0.71 [0.65, 0.76], p ~ 10⁻⁴⁷; (B) Reaction norm plots of the number of branches for the 25 least and most plastic lines. Lines are coloured by the average days to flowering of each accession. (C) Correlation between mean shoot branching plasticity and total number of siliques produced at the senescence stage for 278 accessions. Pearson’s correlations and 95% CI: r(LN) = -0.36 [-0.46, -0.26], p ~ 10⁻¹⁰; r(HN) = 0.41 [0.31, 0.50], p ~ 10⁻¹³ (D) Correlation between mean days to flowering and branching plasticity. The blue line is a smoothed trend fitted by local regression (LOESS). The dotted line shows a cutoff of 25 days to flowering on LN. Pearson’s correlations are shown for all 297 accessions (r = 0.36 [0.26, 0.46], p ~ 10⁻¹⁰) and for 266 early-flowering accessions only (r = 0.45 [0.34, 0.54], p ~ 10⁻¹⁴). In all panels, data are means from n = 4–8 replicates of each accession (median n = 8).

Fig 5. Expression of nitrate-responsive genes in low and high plasticity genotypes.

Transcript levels were assessed by RT-qPCR in pooled seedlings growing in media in which nitrate was replaced by 0.5mM ammonium succinate; after 10 days the seedlings were treated with 5mM KNO₃ or KCl for 2 hours and used for RNA extraction. Previous work has shown that GSR1 is down-regulated in response to nitrate, whereas all the other genes are up-regulated [39,117,118]. The expression of each gene is shown as the log₂(fold-change) between treatment (KNO₃) and control (KCl) conditions (a value of zero indicates no difference between treatments—dashed line). The fold-change in expression was calculated using the “Delta Cp” (ΔΔCp) method [106] and normalised to two reference genes (APX3 and UBC9). Two biological replicates are plotted for each genotype. Genotypes are ordered by their branching plasticity (average branches on HN—LN) from the experiments detailed in Figs 1–4: Sha = 0.6, MAGIC.11 = 1.1, Col-0 = 1.4, Hi-0 = 1.6, Tsu-0 = 3, Rsch-4 = 3.8, MAGIC.345 = 6.4. None of the genes had a significant correlation between gene expression and shoot branching plasticity (in all cases the Spearman’s rank correlation p-value > = 0.2, bonferroni-adjusted for multiple testing).

Fig 6. Grafting experiments between low and high plasticity lines.

Mean shoot branching and days to flowering in reciprocal grafts between low plasticity (Sha accession and MAGIC.11) and high plasticity (Rsch-4 accession and MAGIC.345) lines. Data from two replicate experiments is shown (solid and dashed lines). We’ve tested the hypothesis of no effect of root and shoot genotypes as well as their interaction with the nitrate treatment (plasticity), for each trait using Wald F tests in a mixed model ANOVA (accounting for variation between experiments). For shoot branching there was a significant shoot-by-nitrate interaction [Wald F(1, 9.37) = 77; p ~ 7x10^-6], after accounting for significant marginal effects of nitrate and experiment repeat. For flowering there was no significant organ-by-nitrate interaction (as expected from the lack of plasticity in this trait), but a significant marginal effect of shoot ideotype on the trait [Wald F(1, 9.10) = 264; p ~ 5x10^-8]. There was no detectable effect of root or its interaction with nitrate. Data are means from n = 7–19 replicates per graft in each experiment (median n = 13); error bars are 2x standard error of the mean. Plants were scored at the 2-silique stage. Ungrafted and self-grafted plants are included as controls.

Fig 7

QTL mapping for days to flowering (A) and shoot branching (B-C) in the MAGIC lines. (A-B) QTL scans for each trait under high (HN) and low (LN) nitrate conditions and for shoot branching plasticity. To account for the correlation between shoot branching traits and flowering time, we also carried out the association test using flowering time as a covariate (dotted lines in B). (C) For shoot branching, we also fit a mixed model to the whole dataset simultaneously to separate common genetic (G) effects from gene-by-environment interaction (GxE) effects at each marker (this is somewhat equivalent to the plasticity QTL scan shown in panel B). In all panels, the plots show the LOD score of the association test carried out for each marker along the genome (see methods). The numbered panels correspond to each of the 5 chromosomes of Arabidopsis. The horizontal lines show the 5% genome-wide significance level based on 1000 permutations. Candidate QTL above this threshold are annotated for each trait and condition: SB, shoot branching; FT, flowering time; HN, high nitrate; LN, low nitrate; Pl, plasticity; SB~FT shoot branching including flowering time as a covariate. In panel C they are simply highlighted in red.

Fig 8

QTL mapping for days to flowering (A) and shoot branching (B) in accessions. Manhattan plots showing the association mapping results for each trait on high (HN) and low (LN) nitrate. For shoot branching, QTL mapping was also performed for this trait’s plasticity. The upper horizontal dashed line is the 5% genome-wide significance threshold, obtained with bonferroni correction; one SNP above this threshold is shown in red. The lower line, at p = 10⁻⁵, was defined based on the inclusion of two known QTL for flowering time; SNPs above this relaxed threshold are shown in orange. The association test was carried out using a linear mixed model that corrects for population structure by taking into account the genetic relatedness between individuals [114]. The tests used 192863 bi-allelic SNPs that had >5% frequency in our sample of 240 early flowering accessions scored at 2-silique stage (flowering <25 days on LN).

Fig 9. Predicted parental effects for QTL identified in this work.

The QTL are named as in S4 Fig, Fig 7 and S1 Table. The points and error bars are the mean and 95% confidence intervals of the Best Linear Unbiased Predictors (BLUPs) of the 19 haplotype effects estimated using the R/qtl2 package [43]. The y-axis shows the standardized QTL effect (i.e. the values indicate how many standard deviation units each estimate deviates from the trait’s mean).