spacemodR (development version)

New Features & Improvements

• Species occurrences from GBIF: new function get_gbif_data(), which downloads the occurrence records of a species for a region of interest with the public GBIF search API (no account needed) and returns them as sf points.
• New article “Species Occurrences (GBIF)”, which compares the records of the wood mouse (Apodemus sylvaticus) with its habitat scores from EUNIS and OCS-GE.

spacemodR 0.3.5

New Features & Improvements

• Stacking order of habitat layers: habitat(), add_habitat() and add_nohabitat() gain a priority argument, stored in a new priority column of the habitat object. In habitat_raster(), overlapping features are stacked by priority and the pixel takes the value of the top one, so the result no longer depends on the order of the add_*() calls. By default each new layer goes on top of the previous ones, as before.
• Continuity of thin structures: habitat(), add_habitat() and add_nohabitat() gain a touches argument, stored in a new touches column. With touches = TRUE, lines and polygons fill every pixel they touch in habitat_raster(), which keeps hedgerows, rivers or roads continuous; with touches = FALSE, polygons fill the pixels whose centre they cover.
• Mixed geometries: habitat_raster() now accepts points, lines and polygons in the same habitat object.
• The “Habitat Layer” article explains priority and touches on a watercourse of the Metaleurop landscape.
• OCS-GE from IGN: new function get_ocsge(), a single entry point to get the OCS-GE of a region of interest, with a source argument. source = "ign" calls the new get_ocsge_data_ign(), which downloads the per-department GeoPackages of the IGN Géoplateforme (no API key), keeps them in a cache directory and lets you choose the vintage with year. source = "fgb" calls get_ocsge_data_fgb(), unchanged. Extracting the IGN archives needs the archive package or the 7z command line tool.
• OCS-GE / EUNIS crosswalk: new dataset ref_ocsge_eunis, a ranked correspondence between the 14 OCS-GE cover classes (CS), optionally refined by land use (US), and the 45 EUNIS Level 2 classes, with the share of each match measured on the Metaleurop site.
• Land use in OCS-GE habitats: ocsge_species_dict now has rows for 47 (cover, use) pairs (code_us) in addition to the 14 covers, derived from eunis_species_dict through ref_ocsge_eunis. join_ocsge_species() gains a us_col argument: when the polygons carry the land use, they take the scores of their (cover, use) pair, and of their cover otherwise. New dataset ocsge_us_metaleurop, the OCS-GE of the Metaleurop site with cover and use.
• Habitat data sources guide: the article “Habitat Layer from EUNIS” becomes “Habitat Data Sources”. It compares OCS-GE cover, OCS-GE cover and use, and EUNIS, shows how to move between them, and lists the other sources that could refine them.

Behaviour changes

• touches defaults to TRUE, whereas habitat_raster() used to fill the pixels whose centre is covered by a polygon, and the pixels of a thin line. Existing rasters therefore change: features are drawn larger. Set touches = FALSE to get the previous result, in particular for layers of contiguous polygons such as a land cover map.
• ocsge_species_dict rebuilt with the columns of eunis_species_dict: weight_global becomes suitability, weight_movement becomes permeability, weight_breeding and the species-level columns (sppID, taxon_group, habitat_breadth, specialization, confidence, in_france) are added, and cover codes are written without space ("CS2.2.1"). The scores themselves changed: they follow an anchored scale, are lower on classes that are not habitat, and movement accounts for flight. resistance is now on a 1-100 logarithmic scale derived from permeability (it was a 1-10 score): code that rescaled it, or tested resistance == 10, must be updated (use permeability == 0 for impassable classes).
• nom_espece renamed species_name in ocsge_species_dict and eunis_species_dict.
• plot_species_habitat() now expects suitability instead of weight_global, and draws resistance on a log scale.

Bug fixes

• habitat_raster() no longer fails on an empty habitat object.
• get_eunis_data() returned the raster upside down (north and south swapped), because the image exported by the EEA service carries no georeferencing. The image is now georeferenced before being read. The Metaleurop layer shipped in inst/extdata/eunis_metaleurop_L2.tif was rebuilt; rasters obtained with earlier versions must be downloaded again.

Packaging

• Preparation of the CRAN submission: shorter package title, get_eunis_data() now takes service_url = NULL by default (the EEA service, as before), the examples of habitat_raster() and plot() for habitat objects are run, and the vignette figures are lighter.
• calibrate_simple() checked for the wrong compiled model before sampling.
• dispersal(method = "omniscape") reports its progress with message() instead of cat().

spacemodR 0.3.0

New Features & Improvements

• Soil interpolation (kriging): new functions to build soil contamination rasters from sampling points, without additional dependency: soil_variogram(), fit_soil_variogram(), soil_vgm(), krige_soil() (ordinary kriging or kriging with external drift), cv_soil() (random or spatial-block cross-validation), kernel_smooth_soil() (Gaussian smoothing reusing compute_kernel()), soil_grid() and source_covariates() (distance / wind covariates around a point source). New dataset soil_metaleurop (595 soil Cd, Pb, Zn samples) and new article “Soil contamination maps by kriging”.
• EUNIS land cover: new function get_eunis_data() to download EUNIS Level 1 / Level 2 land cover rasters for a region of interest from the EEA Ecosystem Type Map v3.1 service, and new lookup dataset ref_eunis.
• Trophic graph plots: plot() for trophic objects gains link weights and custom colors.
• Dispersal: cleaned up the kernel dispersal code.

Packaging

• Installable directly from GitHub with remotes::install_github("Qonfluens/spacemodR").
• Source and binary packages are attached to each GitHub release.

spacemodR 0.2.0

Breaking Changes

• Refactored trophic fluxes interface: The intake() function has been completely redesigned and renamed to flux() to better reflect the mathematical and ecological concept of transfer rates.
• New pipe-friendly syntax: The definition of transfer parameters no longer uses multiple named arguments. Users must now initialize the table with flux() and chain specific rules using the new add_flux() function.
• Namespace conflict resolution: The internal utility function previously named flux() (used within the transfer() function) has been renamed to avoid conflicts with the new main API.

New Features & Improvements

• New add_flux() function: Allows for adding transfer rules in a readable, sequential manner (e.g., add_flux(from = "soil", to = "plant", value = ~ 10^x / 32)).
• Global target rules: In add_flux(), if the source compartment (from) is not specified, the rule (value) is automatically applied to all links pointing to the target compartment (to).
• Safe default values: The flux() function now features an explicit default parameter. It is highly recommended to use 1 (full transfer) or 0 (no transfer) instead of NA to prevent the propagation of missing values during subsequent model calculations.

spacemodR 0.1.3

• Bug Fixes:
  • Fixed an rlang_error (replacement has length zero) in spacemodel raster layer sub-assignment within the Tutorial vignette.
  • Removed non-ASCII characters (smart quotes) from the DBFunc_BirdFuncDat dataset to pass CRAN checks.
  • Resolved R CMD Check notes regarding no visible binding for global variable (weight_global, weight_foraging, resistance).
• Documentation & Data:
  • Added complete roxygen2 documentation for internal datasets: FmrBT, DBFunc_MamFuncDat, and DBFunc_BirdFuncDat.
  • Translated and enhanced the Food_Intake vignette into professional scientific English.
  • Updated the pkgdown website structure to use a clean 3-tier navigation system (Quick Start, Core Guides, Example Zoo).

spacemodR 0.1.2

• New Features:
  • Added the Food_Intake module and dataset integration (FmrBT, EltonTraits) to compute Daily Food Intake (DFI).
  • Included the Eco-SSL (Ecological Soil Screening Levels) risk index assessment example.
• Enhancements:
  • Improved the Omniscape_Connectivity tutorial with real-world spatial data for the Wood Mouse (Apodemus sylvaticus).

spacemodR 0.1.0

• Initial Release:
  • Introduced the core spacemodel class for merging raster stacks and trophic networks.
  • Added transfer and dispersal modeling functions.
  • Integrated external Julia Omniscape connectivity algorithms.