Raw data and analysis data for article "Digital herbaria reveal global reorganization of herbivory"

The study examines how herbivory rates vary across latitudes, climate gradients, extreme weather events, ENSO cycles, temporal trends, and protection status using phylogenetic generalized linear mixed models (PGLMMs).| File | Description || `erosion_climate.xlsx` | Main dataset containing 49,827 herbivory records from 535 plant species across 10 families, spanning 1951–2024.

Includes herbivory rates, geographic coordinates, collection dates, taxonomic information, and 19 WorldClim bioclimatic variables (bio1–bio19) matched at each sampling location. || `oni.csv` | Oceanic Niño Index (ONI) from NOAA, organized by year and 3-month periods (DJF, JFM, ..., NDJ), used for ENSO classification and lag analysis. || `leaf_habit.csv` | Leaf habit classification (evergreen/deciduous) for 559 species, sourced from TRY, BIEN, and GIFT databases with genus-level inference for unmatched species. |## Analysis Code| File | Description || `run_latitude.R` | Latitudinal gradient analysis using Bayesian PGLMM with species random effects. || `run_climate_bic.R` | Exhaustive BIC search for optimal climate variable selection from 19 WorldClim variables. || `run_extreme.R` | Extreme weather effects analysis (hot, cold, wet, dry days) on herbivory rates. || `run_enso_sem.R` | ENSO structural equation model (SEM) with PCA-based climate mediation and multi-lag sensitivity analysis. || `run_temporal.R` | Historical temporal trend analysis of global herbivory rates (1951–2024). || `run_protection.R` | Protected vs. non-protected area comparison with spatial grid random effects. || `run_hemisphere.R` | Separate latitudinal gradient models for Northern and Southern hemispheres. || `run_leaf_habit.R` | Latitude × leaf habit (evergreen/deciduous) interaction analysis. || `run_enso_gam.R` | ENSO–herbivory GAM analysis with AR1 autocorrelation correction. || `query_leaf_habit.R` | Script to query TRY/BIEN/GIFT databases for leaf habit data. |## Intermediate ResultsThe `results/` directory contains CSV files with model outputs including conditional effects, model coefficients, BIC comparisons, and effect decompositions for each analysis.