Changes in version 1.2.0 o Expanded documentation for all exported functions, fitted-result classes, model outputs, and interpretation o Revised all six vignettes to use the current API and provide clearer workflow guidance and result interpretation o Added informative print, summary, and response-plot methods for single-surface and multisurface optimization results o Reworked Diagnostic.Plots() to support vector, matrix, dist, and file inputs and to return the fitted model and plotted data o Corrected Resist.boot() to calculate holdout RMSE and added reproducible Windows-compatible parallel bootstrap execution o Added explicit bootstrap parallel controls to all_comb() and protected existing result directories from silent deletion o Hardened Windows parallel optimization by safely transferring SpatRaster inputs to workers and added regression coverage for single-surface, multisurface, and bootstrap workflows o Improved Julia discovery, environment restoration, Circuitscape loading, and current-map handling with live Julia integration tests o Reconciled optional dependencies and namespace imports for package-check compliance o Added CRAN submission metadata and losslessly compressed bundled fitted-result objects to reduce package size o Updated fixed-effect formula extraction for compatibility with current lme4 and reformulas releases Changes in version 1.1.23 o Corrected the Beninde et al. 2024 reference title in the ResistanceGA and Surface_Transformations vignettes o Replaced an incorrect Row et al. 2010 reference entry with the correct citation in the ResistanceGA vignette o Removed a misattributed Beninde et al. 2024 citation from the R-squared objective-function guidance in the ResistanceGA vignette o Verified all remaining vignette and documentation references against published sources Changes in version 1.1.22 o Added a package hex logo (man/figures/logo.png), displayed in the README and on the package help page Changes in version 1.1.21 o Corrected the GettingStarted vignette argument table defaults for run (25), parallel (FALSE), and pop.size (auto-computed) o Fixed the README description of rga2_demo_covariates to list all six raster layers (cti, err27, ffp, gsp, hli, nlcd) o Corrected the Beninde et al. 2024 reference title in the GettingStarted vignette o Added a Julia and JuliaConnectoR version-compatibility note to the Julia_Guide vignette, including a Windows Store alias caveat Changes in version 1.1.19 o Routed jl.prep kept-pair outputs more directly through the shared MLPE pair-metadata workflow while preserving legacy return objects o Added benchmark-based recommend_commute_settings() guidance for choosing exact versus aggregate commute-distance settings o Added focused regression coverage for jl.prep pop2ind/pairs_to_include alignment and commute recommendation thresholds o Updated README and workflow vignettes to demonstrate recommend_commute_settings() for choosing commute-distance optimization settings o Added graph_to_keep() to convert population graphs, adjacency matrices, or edge lists into pair-pruning vectors for gdist.prep() and jl.prep() o Documented graph-based pair pruning in the gdistance and Julia prep helpers o Added min.max_dist support in gdist.prep() using the selected-pair keep-vector workflow o Expanded jl.prep() population-level covariates for pop2ind workflows before applying selected-pair filters o Added a local-only future_dev_ideas.md planning note to .gitignore so future methodological ideas can be kept in the repo workspace without being tracked Changes in version 1.1.18 o Refactored jl.prep to build Julia MLPE workflow data, random-effect terms, and kept-pair subsets through the shared pair-metadata path while preserving legacy return values o Added focused regression coverage for jl.prep pairs_to_include, pop2ind, and nb combinations Changes in version 1.1.17 o Routed all MLPE.lmm_coef coefficient-extraction branches through the dyad-aware MLPE backend while preserving legacy file-output workflows o Added focused regression coverage for file-based MLPE coefficient extraction metadata reuse and formula-driven predictor refresh Changes in version 1.1.16 o Stabilized the categorical optimizer regression test against platform-specific non-finite likelihoods from near-perfect fixture data Changes in version 1.1.15 o Pre-aggregated resistance stacks for opt-in aggregate gdistance commute-distance GA fitness evaluation while keeping final reported gdistance distances exact o Added focused helper-level tests for approximate gdistance optimization setup without running full GA searches Changes in version 1.1.14 o Optimized gdistance commute runs with batched sparse solves and selected-pair gdistance runs by reusing endpoint distance calculations o Added a vectorized sparse transition builder for the default gdistance transition function on projected 4- and 8-neighbor rasters o Added a vectorized projected-raster geographic correction path for gdistance transition matrices o Tuned the sparse Cholesky strategy used by fast gdistance commute-distance solves on large rasters o Reduced peak memory and solve time for exact gdistance commute-distance runs by solving endpoint right-hand sides in blocks o Cached projected-raster gdistance transition geometry across repeated calls with identical raster geometry and directions o Added an opt-in aggregate approximation for iterative gdistance commute-distance optimization runs o Added benchmark scripts comparing optimized gdistance commute runs with Julia/Circuitscape and aggregate approximations Changes in version 1.1.13 o Made the Julia_Guide connection-verification chunks display-only so vignette rebuilds do not depend on starting Julia during package checks Changes in version 1.1.12 o Made the Julia_Guide vignette detect Julia automatically so vignette rebuilds do not fail on systems without a local Julia installation o Restored normal vignette builds in the GitHub Actions package-check workflow Changes in version 1.1.11 o Excluded GitHub workflow files from built package contents and aligned CI package checks with the standard no-vignette pre-check workflow Changes in version 1.1.10 o Updated the GitHub Actions package-check workflow to skip vignette builds on CI runners Changes in version 1.1.9 o Updated package examples and vignettes to use packed-raster-safe resistance surface handling o Fixed jl.prep's internal Julia test raster export to work with the current bundled surfaces o Added a GitHub Actions workflow that runs package checks on substantive pushes and pull requests Changes in version 1.1.8 o Routed bootstrap, diagnostic, and legacy coefficient MLPE fits through the dyad-aware fitting backend o Preserved legacy MLPE.lmm numerical behavior while aligning workflow pair metadata with the new MLPE machinery Changes in version 1.1.7 o Clarified that mlpe and mlpe_data are the preferred standalone MLPE API o Marked mlpe_rga, MLPE.lmm, and To.From.ID as legacy compatibility helpers in documentation Changes in version 1.1.6 o Fixed recursive printing of bootstrap tables in all_comb and bootstrap summaries Changes in version 1.1.5 o Fixed all_comb bootstrap metadata alignment when Null models are included Changes in version 1.1.4 o Integrated the dyad-aware MLPE interface and pairwise environmental helper workflow into main o Updated IBE/IBR and Julia vignettes for the new MLPE formulation and packed terra demo objects o Fixed Plot.trans duplicate plotting, Run_gdistance example behavior, and in-memory current-map returns Changes in version 1.1.2 o Fixed Windows parallel optimization for raster-based workflows o Added IBE/IBR vignette builder support for multi-core execution Changes in version 1.1.1 o Expanded unit tests o Improved error catching Changes in version 1.1.0 o Working package with unit tests Changes in version 1.0.1 o Vignette update Changes in version 1.0.0 o Initial commit of the updated package