Data from: Trait variation and performance across varying levels of drought stress in cultivated sunflower (Helianthus annuus L.)

Earley, Ashley1 ; Nolting, Kristen1 ; Donovan, Lisa2 ; Burke, John1

Research facility: University of Georgia

Published Jun 05, 2024 on Dryad. https://doi.org/10.5061/dryad.xgxd254np

Data files

Jun 05, 2024 version files 5.25 GB

Abstract

Background and Aims: Drought is a major agricultural challenge that is expected to worsen with climate change. A better understanding of drought responses has the potential to inform efforts to breed more tolerant plants. We assessed leaf trait variation and covariation in cultivated sunflower (Helianthus annuus L.) in response to water limitation.

Methods: Plants were grown under four levels of water availability and assessed for environmentally induced plasticity in leaf stomatal and vein traits as well as biomass (performance indicator), mass fractions, leaf area, leaf mass per area, and chlorophyll content.

Key Results: Overall, biomass declined in response to stress; these changes were accompanied by responses in leaf-level traits including decreased leaf area and stomatal size, and increased stomatal and vein density. The magnitude of trait responses increased with stress severity and relative plasticity of smaller-scale leaf anatomical traits was less than that of larger-scale traits related to construction and growth. Across treatments, where phenotypic plasticity was observed, stomatal density was negatively correlated with stomatal size and positively correlated with minor vein density, but the correlations did not hold up within treatments. Four leaf traits previously shown to reflect major axes of variation in a large sunflower diversity panel under well-watered conditions (i.e., stomatal density, stomatal pore length, vein density, and leaf mass per area) predicted a surprisingly large amount of the variation in biomass across treatments, but trait associations with biomass differed within treatments. Additionally, the importance of these traits in predicting variation in biomass is mediated, at least in part, through leaf size.

Conclusions: Our results demonstrate the importance of leaf anatomical traits in mediating drought responses in sunflower, and highlight the role that phenotypic plasticity and multi-trait phenotypes can play in predicting productivity under complex abiotic stresses like drought.

README: Leaf traits predict performance under varying levels of drought stress in cultivated sunflower (Helianthus annuus L.)

Journal: AoB Plants

Corresponding author: John Burke - jmburke@uga.edu

Description of the data and file structure

Experimental Design.csv
File listing genotypes (SAM lines), blocks and treatments for all 64 plants

TraitInfo.xlsx
Table listing all traits and units (also Table 1 in manuscript)

*Code_and_Output -> Data

DataAvg.csv
Plant level data for all traits (see file TraitInfo.xlsx for traits and units)

DataAvg_2.csv
Processed data (processing shown in script 01_Data_Wrangling.R)

Watering_BeforeAfter_Per.csv
Table with percentages of field capacity for watering each day (before and after watering)
RDPI_for_plots.csv
Plant level data used for plasticity (RDPI) analyses in manuscript Figure 2

Code_and_Output -> Code
*R scripts used for all analyses

01_Data_Wrangling.R

Reformatting dataframe to be used in downstream analysis and figures

02_ANOVAs.R

Two-way ANOVAs with interaction on ranked scale data to determine any significant treatment, genotype, and treatment x genotype interactions. Output reported in Tables 2 and S1, and used to determine posthoc treatment differences in Figure 1.

03_Boxplots.R

Code for creating Figure 1, boxplots on a subset of the traits presented in Table 1. Treatment differences follow ANOVA results in Tables 2 and S1.

04_RDPI_plotting.R

Code for creating Figure 2, depicting RDPI (i.e., plasticity) for traits in each drought treatment relative to the well-watered control. Traits are grouped into functional groupings following Table 1 (each trait group was saved individually, and collated into a single figure using a graphics editor). See main text for how RDPI was calculated using estimated genotypic means (marginal means) for each treatment.

05_Corr_Mats.R

Code for creating Figure 3B (correlation matrix across all treatments) and Figures 4B,D,F,H (correlation matrices within each treatment) using all traits from Table 1, except using averages for stomatal density, stomatal pore length, and stomatal guard cell width. Additionally, code for mantel tests comparing each within treatment correlation matrix against each other.

06_PCAs.R

Code for creating Figure 3A (PCA across all treatments; first 2 PC axes) and Figures 4A,C,E,G (PCAs within each treatment; first 2 PC axes) using all traits from Table 1, except using averages for stomatal density, stomatal pore length, and stomatal guard cell width (with trait loadings for PCs 1-3 reported in Table S2). Additionally, code for Hotelling's t tests comparing each within treatment against each other in Figure 3A, using the first 2 PC axes.

07_Trait_Performance_Regressions.R

Code to perform the trait-performance multiple regression analyses using the Bayesian package 'brms'. See main text for details on model fits. Additionally, code to create Figure 5 and Tables S3 & S4 (which report estimated associations for each trait and biomass within each treatment along with credible intervals), and S5 which reports the model comparison results from the models with and without leaf area.

08_WateringData_Plot.R

Code for creating Supplemental Figure 1, showing percent water content across the experiment across all treatments.

*Code_and_Output -> Output
*mod_brm_1 and mod_brm_2: Saved brms model files

*StomataDensity_Raw
Images taken to measure stomatal density.
File Names = -, ex: 1-B1 for pot 1 and bottom stomata density image 1

Scale: 20X, 5.87 px/um

*IndividualStomata_Raw
Includes images of individual stomata used for stomatal size measurements. Images were taken
on two difference microscopes, designated 1 and 2

File names = -S, ex: 1-BS1 is Pot 1, Bottom Stomata image 1

Microscope 1 Scale: 100X, 29.45 px/um
Microscope 2 Scale: 40X, 15.96 px/um

*LeafScans_Raw
Images of scans taken on flatbed scanner at 300dpi for measurement of LMA and leaf area
File Names = Pilot_, ex: Pilot1_1 for pot 1, image 1

*MinorVeins_Raw
Images taken with light microscope and used in neural network to calculate VLA.

File Names = -, ex: 1-1 is pot 1, image 1

Microscope 1 Scale: 5X, 1.4688 px/um

*MajorVeinsScans_Raw
Images taken on flatbed scanner at 2400 dpi for measurement of second and major veins

Note:
**Any missing values (NAs) in the dataframes indicate either the plant died during the experiment or accurate measures could not be made from the images taken.

Code and Software

Analyses were conducted using R v43.24.13 and RStudio v1.3.1093. All scripts are in folder Code_and_Output -> Code