Research compendium to reproduce analyses and figures of the article: "Direct and trait-mediated effects of climate on grassland productivity across the Alps"

Reproducibility instructions Overview This repository contains all scripts required to reproduce the data processing and statistical analyses presented in the manuscript. The workflow is organized in sequential steps, from remote sensing data extraction in Google Earth Engine (GEE) to statistical modelling in R. The complete workflow requires: A Google Earth Engine account R (≥ 4.x) Required R packages listed in the R scripts All analyses were conducted using R. Step 1 – Remote sensing data extraction (Google Earth Engine) All raw data used in this study are publicly available and were accessed and processed in Google Earth Engine (GEE). The spatial mask defining natural alpine grasslands above 1,000 m a.s.l. across the European Alps is available as a GEE asset at:https://code.earthengine.google.com/?asset=users/pierregauzere/natural_grasslands_1000m_europe The set of 10,000 randomly selected grassland plots (60 m radius buffers) used as units of analysis is available as a GEE vector asset at:https://code.earthengine.google.com/?asset=users/pierregauzere/NatGrass_Alps_buffers60m_10000 Sentinel-2 Level-2A surface reflectance imagery (COPERNICUS/S2_SR_HARMONIZED), climate reanalysis products (ERA5-Land and TerraClimate), and static environmental layers (Copernicus DEM, WorldClim, OpenLandMap, HydroATLAS) are all publicly available datasets accessed directly within GEE. All processing steps used to derive plot-level spectral traits (CIre, CCI, NDII), vegetation productivity (NIRv), interannual climate anomalies (GDD, CWD, SWE), and static environmental variables are fully documented and reproducible via the GEE scripts provided by the authors. The main scripts can be accessed and executed at the following links: Script for computing spectral traits and productivity from Sentinel-2 imagery:https://code.earthengine.google.com/045dff0eec966c841f443feb4bd1eb7c?accept_repo=users%2Fgisizw%2FGEE-PICX Script for computing interannual climate time series (GDD, CWD, SWE):https://code.earthengine.google.com/9e8314e07b23d0fab6fcc8eb3aaf83d2?accept_repo=users%2Fgisizw%2FGEE-PICX Script for compiling static environmental variables:https://code.earthengine.google.com/9e8314e07b23d0fab6fcc8eb3aaf83d2?accept_repo=users%2Fgisizw%2FGEE-PICX To reproduce this step: Open each script in the GEE Code Editor. Run the scripts. Export all generated outputs (tables and/or rasters) to Google Drive. Step 2 – Export GEE outputs locally Download all exported files from Google Drive and place them in the following local directory: data/GEE_Export/ The folder structure must be preserved exactly as expected by the R scripts. No renaming of files should be performed unless explicitly indicated in the scripts. Step 3 – Preparation of vegetation indices Run the following script: analyses/prepare_vegetation_indices.R This script: Cleans and harmonizes GEE exports Computes derived vegetation indices Formats time series Produces intermediate processed datasets Output files are saved in the appropriate code outputs/ subdirectory. Step 4 – Creation of the full analytical dataset Run: analyses/create_full_dataset_grasslands_10000.R This script: Merges vegetation indices, climate predictors, and ancillary variables Applies filtering criteria described in the manuscript Constructs the final analytical dataset used in statistical models The resulting dataset is saved in: code outputs/grasslands_full_dataset.RData Step 5 – Main statistical analyses Finally, run: analyses/full_analyses.R This script reproduces: Descriptive analyses Model fitting Structural equation models (if applicable) Figures and tables presented in the manuscript All outputs are written to: code outputs/ figures/ Reproducibility notes All scripts are designed to be run sequentially in the order described above. Paths are relative to the root directory of the repository. Random seeds are set internally where stochastic procedures are involved.