Aridity and grazing are associated with reduced trait complementarity and higher invasion intensity of Solanum rostratum in native plant communities
Data files
Sep 15, 2025 version files 1.02 MB
-
code.R
13.22 KB
-
data.xlsx
744.93 KB
-
LMM_Interaction_Results_Full.csv
25.64 KB
-
LMMdata.csv
98.06 KB
-
LMMResults.csv
11.24 KB
-
Muldata.csv
18.96 KB
-
pcadata.csv
79.57 KB
-
random_factors.csv
1.77 KB
-
README.md
18.15 KB
-
SEMdata.csv
9.87 KB
Abstract
1The biotic resistance hypothesis (BRH) of Elton posits that diverse communities are more resistant to biological invasions. While the effects of climatic stresses and human disturbances on community invasibility have been extensively studied individually, their combined and potentially interactive influences remain poorly understood.
To address this problem, a national-scale survey was conducted on 3000 km in China to assess the relationship between the intensity of Solanum rostratum invasion and the diversity of native species.
Our study found that sites with higher native plant biodiversity generally exhibited lower S. rostratum invasion intensity. Specifically, native plant diversity helped resist invasion by improving community complementarity, increasing community density, coverage, and biomass, promoting community weighted means (CWM) of resource-conservative traits, and reducing trait differences between invasive and native plants. Furthermore, biodiversity loss was associated with higher S. rostratum invasion intensity. Specifically, sites with higher aridity and grazing tended to have lower biodiversity, reduced community complementarity, decreased density, coverage, and biomass, and lower community-weighted means of resource-conservative traits. In particular, phylogenetic diversity (Faith’s PD) and the Simpson index were more effective than species richness in predicting the resistance of local communities to invasion by S. rostratum and showed stronger negative correlations with invasion intensity.
Our results further supported the BRH and emphasized the importance of considering species richness, evenness, phylogenetic structure, and trait structure when explaining biological resistance to invasion. Overall, this study highlighted the crucial role of the diversity and structure of the native plant community in resisting S. rostratum invasion. Sites experiencing higher aridity and grazing were associated with reduced resistance to invasion, as indicated by lower biodiversity and reduced community complementarity*.* Therefore, conserving and restoring native plant diversity, particularly enhancing phylogenetic diversity and resource-conservative traits, can improve the resistance of the ecosystem to invasive species.
Dataset DOI: 10.5061/dryad.v6wwpzh85
Description of the data and file structure
This document describes the contents of the data files and analysis code associated with the manuscript: "Aridity and grazing weaken the biotic resistance of native plant communities to Solanum rostratum’s invasion by reducing trait complementarity and resource-conservative traits"
Files and variables
File: code.R
Description: This R script contains all the statistical analyses presented in the manuscript. It integrates all major components of the data analysis workflow, including:
Linear Mixed-Effects Models (LMMs):
Assessing associations between ecological predictors and invasion intensity across multiple response variables, with pairwise interaction screening and automated extraction of model outputs.
Principal Component Analysis (PCA):
Dimension reduction of community-weighted mean traits, dissimilarity metrics, biodiversity indices, and ecosystem structure variables for multivariate visualization and SEM inputs (e.g., Figures S3–S5).
Partial Correlation Analyses:
Testing associations between trait-based and phylogenetic dissimilarity metrics and invasion intensity, while controlling for environmental or community-level covariates.
Structural Equation Modeling (SEM):
Evaluating direct and indirect pathways linking grazing, aridity, community structure, trait complementarity, and invasion intensity using latent constructs and composite variables (e.g., Div_PC1, CWM_PC1).
File: data.xlsx
Description: This Excel workbook contains all raw and processed data used in the analyses presented in the manuscript.
Variables
| Variable | Definition | Unit / Scale | Notes |
|---|---|---|---|
| Site | Sampling site identifier | Categorical | Random effect in LMMs |
| Sample | Sample ID | Categorical | – |
| Lon | Longitude | ° | – |
| Lat | Latitude | ° | – |
| III_SR | Invasion intensity index of S. rostratum | Unitless index (0–1) | Higher values = stronger invasion |
| Grazing | Grazing | Ordinal (0 = none, 1 = grazing) | Assessed by local records & field observation |
| Aridity index (AI) | aridity index | Unitless (0-1) | - |
| Aridity | 1-aridity index | Unitless (0–1) | Higher = drier |
| MAT | Mean annual temperature | °C | – |
| Species_richness | Number of plant species per plot | Count | – |
| Simpson index | Simpson diversity index | Unitless (0–1) | Higher = greater diversity |
| PD | Faith’s phylogenetic diversity | Branch length units | – |
| Plant_density | Plant individual density | individuals·m⁻² | – |
| Vegetation_coverage | Vegetation cover | % | – |
| Biomass | Aboveground plant biomass | g·m⁻² (dry weight) | – |
| MPD_phy | Mean pairwise distance based on phylogenetic distance | Branch length units | – |
| MNTD_phy | Mean-nearest-taxon-distance based on phylogenetic distance | Branch length units | – |
| NRI | Nearest relative index | Unitless (standardized effect size) | – |
| NTI | Nearest taxon index | Unitless (standardized effect size) | – |
| MPD_traits | Mean pairwise distance based on plant trait distance | Branch length units | Gower's distance based on 9 traits |
| MNTD_traits | Mean-nearest-taxon-distance based on plant trait distance | Branch length units | Gower's distance based on 10 traits |
| FDis_all | Functional dispersion index based on 9 traits | Unitless | Higher = more dispersed traits |
| CWM_PC1 | First PCA axis generated by the nine community-weighted meantraits | Unitless (PCA score) | – |
| CWM_PC2 | Second PCA axis generated by the nine community-weighted meantraits | Unitless (PCA score) | – |
| βMPD_phy | Mean phylogenetic distance between the S. rostratum and all other native species in a sampling plot community | Unitless | – |
| βMNTD_phy | Phylogenetic distance between the S. rostratum and nearest native species in a sampling plot community | Unitless | – |
| βMPD_traits | Mean functional distance between the S. rostratum and all other native species in a sampling plot community | Unitless | – |
| βMNTD_traits | Functional distance between the S. rostratum and nearest native species in a sampling plot community | Unitless | – |
| β_PC1 | First PCA axis generated by the nine absolute value of S. rostratum to community-weighted mean traits | Unitless (PCA score) | – |
| β_PC2 | Second PCA axis generated by the nine absolute value of S. rostratum to community-weighted mean traits | Unitless (PCA score) | – |
| FDis_* | Functional dispersion index based on specific trait | Trait-specific units | Gower's distance based on specific trait |
| CWM_* | Community-weighted mean of functional traits | Trait-specific units | Weighted by species relative abundance |
| Delta_*_CWM | Dissimilarity between invader trait and community mean | Trait-specific units | Absolute difference |
| Dom_* | Trait value of dominant native species | Trait-specific units | – |
| Delta_*_Dom | Dissimilarity between invader trait and dominant species | Trait-specific units | Absolute difference |
| Height | Plant height | cm | Measured at peak growth |
| SLA | Specific leaf area | mm²·mg⁻¹ | Area per unit dry mass |
| LDMC | Leaf dry matter content | % | Leaf dry mass / fresh mass × 100 |
| RDMC | Root dry matter content | % | Root dry mass / fresh mass × 100 |
| SRL | Specific root length | cm/g | Root length per unit dry mass |
| Leaf C:N | Leaf carbon-to-nitrogen ratio | – | – |
| Leaf C:P | Leaf carbon-to-phosphorus ratio | – | – |
| Leaf N:P | Leaf nitrogen-to-phosphorus ratio | – | – |
| Leaf K | Leaf potassium concentration | g/kg | – |
File: LMM_Interaction_Results_Full.csv
Description: Contains the full results of linear mixed-effects models (LMMs) including interactions between predictors for multiple response variables.
Variables
- Response: Name of the response variable
- Predictor1: First predictor in the interaction term
- Predictor2: Second predictor in the interaction term
- Term: Fixed effect term: Predictor1, Predictor2, or their interaction
- Estimate: Standardized coefficient estimate
- Std_Error: Standard error of the estimate
- P_Value: p-value of the term
- X95._CI_Lower: Lower bound of 95% confidence interval
- X95._CI_Upper: Upper bound of 95% confidence interval
- Marginal_R_Squared: Variance explained by fixed effects
- Conditional_R_Squared:Variance explained by fixed + random effects
- AIC: Akaike information criterion for model fit
File: LMMdata.csv
Description: Contains predictor and response variables used in LMM analyses. Includes environmental, biodiversity, functional traits, and community dissimilarity metrics.
Variables:
All variables are defined in the data.xlsx section.
File: LMMResults.csv
Description: Results of linear mixed-effects models without interaction terms.
Variables
- Response: Name of the response variable
- Predictor: Name of the predictor variable
- Coefficient: Standardized coefficient estimate
- Std_Error: Standard error of the estimate
- P_Value: p-value of the term
- X95._CI_Lower: Lower bound of 95% confidence interval
- X95._CI_Upper: Upper bound of 95% confidence interval
- Marginal_R_Squared: Variance explained by fixed effects
- Conditional_R_Squared:Variance explained by fixed + random effects
- AIC: Akaike information criterion for model fit
File: Muldata.csv
Description: Dataset for multiple regression and relative importance analyses of functional and biodiversity predictors on invasion.
Variables
Subset of data.xlsx variables: III_SR, Species_richness, PD, Simpson_index, Biomass, Plant_density, Vegetation_coverage, MPD_phy, MNTD_phy, MPD_traits, FDis_all, CWM_PC1, betaMNTD_phy, beta_PC1
File: pcadata.csv
Description: Data for PCA analyses of community-weighted mean traits, trait differences between invader and resident communities, dominant species traits, biodiversity, ecosystem structure, and dissimilarity metrics.
Variables
Subset of data.xlsx variables (CWM_, Delta_CWM, Dom, Delta__Dom, III_SR, PD, etc.)
File: random_factors.csv
Description: Random effect identifiers for LMM and SEM analyses.
Variables
- Site: Sampling site
- random effect: Factor used as random effect in mixed models
File: SEMdata.csv
Description: Data used in piecewise SEM analyses linking environmental factors, biodiversity, functional traits, community structure, and invasion intensity.
Variables
- III_SR: Invasion intensity index of S. rostratum, Unitless index (0–1)
- Grazing: Ordinal (0 = none, 1 = grazing)
- Aridity: (1-aridity index),Unitless (0–1)
- Div_PC1: First PCA axis of diversity indices (including: species richness, PD, Simpson index)
- CD_PC1: First PCA axis of Complementarity\dispersion (including: MPD_phy, MNTD_phy, NRI, NTI, MPD_traits, MNTD_traits, FDis_all)
- ES_PC1: First PCA axis of Ecosystem structures (including: Biomass, Vegetation_coverage, Plant_density)
- CWM_PC1: First PCA axis generated by the nine community-weighted mean traits (including:Height, SLA, LDMC, RDMC, SRL, Leaf C:N, Leaf C:P, Leaf N:P, Leaf K)
- Dis_PC1: First PCA axis of Dissimilarity_PC1 (S. rostratum vs resident species) (including: βMPD_phy, βMNTD_phy, βMPD_traits, βMNTD_traits, β_PC1, β_PC2)
Code/software
Data analyses were conducted using R version 4.2.3 (https://cran.r-project.org), an open-source statistical computing environment. All analysis scripts are provided in the file code.R accompanying this submission.
The analysis workflow includes fitting linear mixed-effects models (LMM), principal component analysis (PCA), partial correlation analyses, and structural equation modeling (SEM). All R packages used are freely available through CRAN or GitHub.
R Packages Used
Linear Mixed-Effects Models and Model Evaluation
lme4 (mixed-effects model fitting)
lmerTest (p-values for mixed models)
MuMIn (model selection, R² calculation)
car (ANOVA, VIF, diagnostics)
broom.mixed (tidy model outputs)
sjPlot (model visualization)
parameters (parameter extraction and summaries)
nlme (alternative mixed model fitting)
QuantPsyc (standardized coefficients, path analysis)
Data Wrangling
dplyr, tidyr (data manipulation and reshaping)
PCA and Multivariate Analyses
FactoMineR (multivariate analysis including PCA)
factoextra (visualization of PCA)
ade4 (ecological multivariate analyses)
Structural Equation Modeling
piecewiseSEM (piecewise SEM for ecological path models)
Workflow Summary
Data preparation and cleaning were performed using dplyr and tidyr, including manual standardization of variables.
LMMs were fit using lme4 and lmerTest, with model summaries extracted and visualized using MuMIn, broom.mixed, and sjPlot.
PCA analyses on trait, diversity, and ecosystem structure datasets were conducted with FactoMineR and visualized using factoextra.
Partial correlation analyses utilized functions from the car package and base R.
SEM was modeled using piecewiseSEM, incorporating mixed and composite variables (e.g., PC1 scores) as latent constructs.
Final outputs, including statistical results and model diagnostics, were exported as .csv files and incorporated into manuscript figures and tables.
Access information
Other publicly accessible locations of the data:
- NA
Data was derived from the following sources:
- NA