Data and code for: "Spreading across the territory: new records and invasion risk assessment of Stenochrus portoricensis Chamberlin, 1922 (Schizomida: Hubbardiidae) in Brazil"

This repository contains all data, R scripts, and model outputs associated with the study reporting new occurrence records of the exotic schizomid Stenochrus portoricensis Chamberlin, 1922 in Brazil and assessing its invasion risk through ecological niche modeling. Occurrence data comprise 171 georeferenced records compiled from published literature, zoological collections, and the World Schizomida Catalog, covering the known global distribution of the species. Records from climate-controlled environments (greenhouses, botanical gardens) were excluded prior to modeling. The dataset also includes occurrence records of native Brazilian Schizomida species obtained from the World Schizomida Catalog (2025), used to assess potential spatial overlap with the exotic species. R scripts cover the complete modeling workflow in sequential order: Spatial thinning of occurrence records (spThin) Definition of the calibration area M as a 200-km buffer around occurrence records (sf) Extraction and multicollinearity screening of bioclimatic predictors, including exclusion of variables BIO-8, BIO-9, BIO-18, and BIO-19 prior to VIF analysis (usdm) Hyperparameter tuning of MaxEnt models across 30 candidate configurations (6 regularization multiplier values × 5 feature class sets) using spatial block cross-validation (ENMeval v2.0) Retreining of the final model with optimized parameters (LQ feature classes, regularization multiplier = 0.5) using the predicts package Consensus projection of future habitat suitability across four CMIP6 General Circulation Models (HadGEM3-GC31-LL, MIROC6, CNRM-CM6-1, MRI-ESM2-0) under SSP1-2.6 and SSP5-8.5 scenarios (2061–2080), retaining only grid cells where at least two of four models agreed (terra, geodata) Mobility-Oriented Parity (MOP) analysis to identify regions of extrapolation risk in future projections (mop) Calculation of the 10th percentile threshold from training presences and generation of binary suitability maps Model performance evaluation using the Continuous Boyce Index (ecospat) Model outputs include current and future habitat suitability rasters in GeoTIFF format, consensus maps for SSP1-2.6 and SSP5-8.5 scenarios showing the number of GCMs predicting suitable conditions per grid cell (0–4), binary suitability maps based on the 10th percentile threshold, and MOP proportion rasters indicating extrapolation risk under each future scenario. All analyses were conducted in R version 4.4.1. Spatial visualization and final map production were performed in QGIS version 3.36. Bioclimatic variables were obtained from WorldClim v2.1 at 2.5 arc-minute resolution (Fick & Hijmans, 2017). Future climate layers were obtained from the WorldClim CMIP6 framework via the geodata package.