Data from: Climate change-driven shifts in C3 and C4 grass distributions and leaf traits could lead to changes in community-level flammability
Abstract
Climate change poses challenges to the grasslands of the North American Great Plains Region (GPR), where shifts in species distributions and fire dynamics are expected. The differential responses of C4 and C3 grass species to future climate conditions, particularly in habitat suitability and flammability, are critical for understanding ecosystem changes. This study uses species distribution models to predict shifts in habitat suitability for 37 species under future climate scenarios and assesses flammability traits in a Free-Air CO2 Enrichment study, focusing on species' physiological responses to elevated CO2, warming, and drought. Our models predict that C4 species will retain higher habitat suitability, while C3 species will decline. Leaf-level flammability analysis shows that species with higher water-use efficiency under elevated CO2 will exhibit reduced flammability, potentially decreasing the predicted rate of spread (ROS) when such species dominate. In contrast, species with higher growth rates but lower water-use efficiency may experience increased flammability. Species-specific responses varied within functional types. Anticipated shifts in species distributions suggest C4 species will become more dominant, potentially altering competitive dynamics and reducing C3 diversity. Changes in flammability under future conditions are expected to influence fire regimes, with a predicted decrease in mean community ROS due to the dominance of less flammable C4 species. These findings highlight the need for adaptive fire management and conservation strategies to maintain biodiversity and ecosystem function in the GPR under climate change.
Raubenheimer, S.L., Zheng, L., Stefanski, A., Reich, P.B.
Citation
Please cite this data and code as:
Raubenheimer, S.L. et al. (2025). Climate change-driven shifts in C3 and C4 grass distributions and leaf traits could lead to changes in community-level flammability. Dryad Digital Repository. https://doi.org/10.5061/dryad.dv41ns29n
Project Description
This repository includes species distribution models (SDMs), trait-based flammability assessments, and fire behavior simulations to explore how climate change may shift species distributions and alter fire regimes in North American grasslands. The study models 37 C3 and C4 grass species under current and future climate scenarios using MaxEnt. Empirical trait data from a Free-Air CO₂ Enrichment experiment (TeRaCON) are used to estimate flammability and fire spread rates with the Rothermel model. Changes in community-level flammability are inferred by integrating predicted species abundances with trait-informed fire behavior models.
Folder and File Descriptions
The dataset consists of the main folder DRYAD.zip, which includes the following files:
Code Files
MAIN CODE_MIROC_GreatPlains_MaxEnt.R
The main analysis pipeline used in the manuscript. Includes:- Species distribution modeling under MIROC SSP585 (2041–2060)
- Habitat suitability projections
- Trait × environment integration
- Species-level and community-level flammability modeling using the Rothermel model
- Supplementary climate scenario SDM scripts
Each replicates the SDM analysis pipeline using a different GCM × SSP scenario (used for sensitivity analyses in the supplementary material):MIROC_370_GreatPlains_MaxEnt.RECEarth_GreatPlains_MaxEnt.REC_Earth_370_GreatPlains_MaxEnt.RCMC_GreatPlains_MaxEnt.RCMC_370_GreatPlains_MaxEnt.RACCESS_GreatPlains_MaxEnt.RACCESS_370_GreatPlains_MaxEnt.R
Data Files and Folders
wc2.1_10m_bio/
WorldClim 2.1 bioclimatic variables at 10 arc-minute resolution, used for ambient (current) SDM projections.2060/
Future bioclimatic predictor layers (2041–2060) for multiple scenarios (e.g., MIROC SSP370/585, ACCESS, CMC, EC-Earth). Structured to match WorldClim format for compatibility with MaxEnt.- The final trait dataset (
data_BioCON_avg) used for fire behavior modeling was generated by merging and processing three source data files collected during the TeRaCON experiment at Cedar Creek, Minnesota (2022–2024). These files are:BioCON_leaf_trait_data_Liting.csv– Leaf morphological and physiological traits measured in 2023, including height, SLA (specific leaf area), and LDMC (leaf dry matter content).Harvest data for SR_240322.csv– Aboveground biomass measurements by species and treatment combination collected in 2023.Sarah Leaf samples.csv– Supplemental leaf trait data collected in 2023, including height, SLA (specific leaf area), LWC (leaf water content), and LDMC (leaf dry matter content) for a subset of species and treatments.
Data Integration
- Trait and biomass data were harmonized by species and treatment (CO₂ × nitrogen × warming × water).
- Species codes were standardized across datasets.
- Functional group (
FGroup) and photosynthetic pathway (Plant_type) were assigned manually based on species identity. - Leaf Water Content (LWC) was derived from LDMC using the formula:
LWC = 100 − LDMC. - Treatments were combined into a categorical variable (
CombinedT) with levels corresponding to simplified combinations of elevated CO₂ (+CO2), nitrogen (+N), warming (+T), and drought (−H2O). - Only eight common grass species (four C3, four C4) were retained for downstream modeling.
- Final dataset (
data_BioCON_avg) summarizes trait and biomass data at the species × treatment level.
Variables in
data_BioCON_avg.csv
| Column Name | Description | Units |
|---|---|---|
Species |
Full species name (e.g., Andropogon gerardii) | – |
Plant_type |
Photosynthetic type of the species (C3 or C4) | – |
CombinedT |
Simplified treatment combination (e.g., +CO2, +N +T, Ambient) |
– |
LWC |
Leaf Water Content, calculated as 100 − LDMC |
percent (%) |
H..cm. |
Vegetative height of the plant | centimeters (cm) |
Biomass.g.m2 |
Aboveground biomass per unit area | grams per m² |
Note: The final dataset only includes treatments used in the main analyses: Ambient, +CO2, +N, +T, -H2O, and +CO2 +N +T -H2O.
- Species occurrence data
Species presence records downloaded and filtered from GBIF using thergbifR package. Data were cleaned for geographic precision and duplicates. - Shapefiles
Boundary shapefiles (e.g.,Great_Plains_LCC_Boundary.shp) used to constrain SDM outputs to the North American Great Plains.
Software Requirements
- R version 4.2.2
- Required R packages:
raster3.6-14,terra1.7-29,dismo1.3-11,sf1.0-13ggplot23.4.0,ggpubr,rasterVis,viridis,wesanderson,stringr,dplyr,tidyr
- MaxEnt version 3.4.4
Java-based MaxEnt software is required to run SDMs.
Download: https://biodiversityinformatics.amnh.org/open_source/maxent/
Place the.jarfile in the working directory or set your system PATH accordingly.
How to Reproduce Main Results
- Open R or RStudio.
- Set the appropriate working directory and paths for your system.
- Run
MAIN CODE_MIROC_GreatPlains_MaxEnt.R. This will:- Generate species habitat suitability models (current and future)
- Summarize results by photosynthetic type (C3 vs C4)
- Project community-level fire spread potential using trait-weighted species distributions
- To explore other climate scenarios, run the corresponding supplementary
.Rscript (e.g.,ACCESS_370_GreatPlains_MaxEnt.R).
Outputs
The code generates:
- Raster maps of habitat suitability per species, per scenario
- Presence/absence and community composition rasters
- Summaries of range shifts by photosynthetic type (C3 vs C4)
- Predicted species-level and community-level fire spread rates
- PNG/PDF plots of spatial predictions
- Summary tables of model performance and trait associations
All output files are saved in the user-specified output_dir.
Notes on Reproducibility
- All environmental predictors and trait data are directly measured or externally sourced and stored in structured folders.
- Random seeds are used for reproducible MaxEnt outputs.
- Code is modular, with scenario-specific
.Rscripts mirroring the structure of the main analysis. - The fire behavior modeling (Rothermel) is integrated directly into the main pipeline and uses species-specific ROS rasters.
Glossary of Key Terms
- SDM: Species Distribution Model
- MaxEnt: Maximum Entropy algorithm used for presence-only species modeling
- GCM: General Circulation Model
- SSP: Shared Socioeconomic Pathway (climate scenario)
- TeRaCON: Terrestrial Responses to Carbon and Nitrogen experiment
- ROS: Rate of Spread (of fire)
- H.S.: Habitat Suitability
License
This dataset and associated code are released under the CC0 1.0 Universal (Public Domain Dedication) license.
Contact the authors for further information or potential collaboration.
Contact
For questions or clarifications, please contact:
Sarah Raubenheimer
Email: sraubs@umich.edu or sarah.l.raubenheimer@gmail.com
Institute for Global Change Biology, University of Michigan