Source data for "More species, fewer specialists: patch size mediates the herb-layer response to lost forest continuity"

This record contains the source data for the research paper “More species, fewer specialists: patch size mediates the herb-layer response to lost forest continuity.” It is the complete, analysis-ready dataset underlying every figure, table and statistic in that paper — a re-analysis of 70 herb-layer plots (0.1 ha) sampled in broadleaved forests of the Zemgale lowland, southern Latvia, spanning a forest-continuity gradient (ancient, old, recent post-agricultural) crossed with forest patch size (large vs. small, isolated fragments). The record deliberately contains only the data used in this paper, so that a reader can reproduce the results without ambiguity. It comprises three tables, a spatial layer, the analysis pipeline and a README/data dictionary: dataset_plots.csv — one row per plot (70): land-use history, patch size, environmental covariates (soil moisture, edge class, canopy closure), and the derived response variables (richness, specialist/generalist counts, specialist share, community mean trait values including myrmecochory, and multivariate dispersion). dataset_species_composition.csv — long format, one row per species × plot occurrence; herb-, shrub- and tree-layer records with Braun–Blanquet cover. Analyses use the herb layer as presence/absence. dataset_species_traits.csv — one row per species: specialist and generalist classifications (forest specialists: ancient-forest indicators, Hermy et al. 1999, and woodland-key-habitat indicators, Ek et al. 2002; disturbance generalists: ruderal and alien species, Gavrilova & Šulcs 1999 and Priedītis 2014) and the functional traits used in the paper. Trait sources: LEDA, TRY, BiolFlor (hemeroby and dispersal syndrome; Kühn et al. 2004), the mycorrhizal-intensity database (Akhmetzhanova et al. 2012) and a national trait compilation. zemgale_study_spatial.gpkg — spatial layers (plots, forest patches, digitized continuity polygons, study-area boundary) in LKS-92 / Latvia TM (EPSG:3059). analysis_pipeline.R — the R script that regenerates every model, ordination, indicator-species analysis and figure in the paper from the files above (stand-level PERMANOVA on forest × continuity units; appendix sensitivity analyses on the herbaceous-species pool; fixed-effects and spatial-extent checks). lofo_specialist_check.R — the leave-one-forest-out sensitivity analysis (Appendix S8). README_dataset.md — full data dictionary and processing notes. Note: the earlier RLQ-formatted trait/species/environment tables (Tables L, Q, R) belong to a separate study and are archived with that paper; they are intentionally not duplicated here, to keep this record specific to the present paper.