Changes in version 0.9.5                        

  - The example dataset is now bombacoideae_occurrences: 17,030
    occurrence records of 185 Bombacoideae species (Zizka et al. 2020),
    the worked example of the companion paper. It is used in the README,
    the vignettes and all function examples; the previous dataset
    biomes_example has been removed.
  - biomes_information gains the column biome_definition (climate,
    vegetation, land_cover, ecoregion, integrative, anthropogenic); the
    definition argument of biomes_rank(), biomes_visualise() and
    biomes_full(scheme = ) ranks within one biome definition. Scheme 6
    (Zhang et al. 2017, climate and NDVI clustering) is integrative.
  - biomes_rank(): the criterion is called effective_biomes (columns
    effective_biomes_raw / _scaled); tiebreaker = "biomes". The ranking
    uses exactly the three criteria of the companion paper (coverage,
    effective_biomes, granularity), min-max rescaled and averaged with
    equal weights; the experimental criteria evenness, informativeness
    and agreement and the scaling argument have been removed. A
    criterion without variation among the compared schemes is left out
    of the composite (attribute criteria_used), and a single compared
    scheme gets no composite score (NA) but is still returned as
    best_scheme.
  - biomes_visualise(): the rank panel labels schemes as 25 (Ramankutty
    & Foley, 1999), shows all compared schemes (or the top_n best) on
    one 0 to 1 axis (the rescaled criterion values that enter the
    composite score; axis titles carry the symbols of the paper,
    Coverage (C), Effective biomes (E), Granularity (G)) and outlines
    the best composite score and the best value of each criterion in
    red; panels carry left-aligned titles with the panel letter and the
    reference of the chosen scheme (titles); the map legend lists biome
    names only (legend_counts = FALSE); the barplot uses the map colours
    per biome, grey for off-map records, and thousands separators.
  - Maintainer e-mail updated.

                 Changes in version 0.9.4 (2026-08-31)                  

  - CRAN resubmission addressing reviewer comments.

                        Changes in version 0.9.3                        

  - Initial CRAN submission.
  - Provides raster layers of 31 global biome schemes from Fischer et
    al. (2022, Global Ecology and Biogeography 31(11): 2172-2183) at 10
    x 10 km resolution globally.
  - Terminology: schemes are addressed by their biome scheme number
    (1-31). The scheme argument of biomes_classify(), biomes_rank(),
    biomes_visualise() and biomes_full() replaces the former layer
    argument; biomes_rank() returns the columns scheme/scheme_name and
    the attribute best_scheme; biomes_tab() returns a scheme column;
    biomes_information uses the column scheme_number.
  - Core functions: biomes_classify() (assign occurrence records to
    biome classes), biomes_rank() (rank schemes by coverage, effective
    number of classes, and granularity), biomes_tab() (tabulate records
    per biome class), biomes_visualise() (combined figure with rank, map
    and barplot panels, selectable via panels), biomes_full() (one-call
    wrapper), biomes_get() (load the raster stack), biomes_info()
    (per-scheme metadata), biomes_occ() (optional GBIF download with
    coordinate cleaning).
  - biomes_visualise() now reproduces the full workflow figure: the rank
    panel shows the composite score plus the raw criteria it averages
    (coverage, effective classes, granularity), the map panel shows the
    occurrence map (with an adaptive legend), and the barplot panel
    shows records and species per biome class back-to-back with centred
    labels. With combine = FALSE the individual panels are returned as a
    named list instead of one lettered figure. The former
    biomes_show_rank() has been removed (its ranking view is the rank
    panel).
  - biomes_full(): the scheme argument also accepts a scheme type
    ("climate", "vegetation", "land_cover", "ecoregion", "integrative",
    "anthropogenic") to pick the best-fitting scheme within that group,
    in addition to an integer 1:31 and "best". A new plot argument
    controls which figure(s) are built: "none" (default, no figure, the
    fastest option), "all" (the combined lettered figure in $plot), or a
    subset of c("rank", "map", "barplot") (returned individually,
    without panel letters, in $rank, $map and $barplot).
  - The ~36 MB biome raster stack is not bundled inside the package. It
    is hosted as a GitHub release asset and downloaded once into a
    per-user cache directory (tools::R_user_dir()) on first use;
    biomes_download() performs (or refreshes) this download explicitly.
    This keeps the installed package well under CRAN's size limit.
  - Four vignettes follow the four-step workflow (step1- .. step4-): (1)
    assembling occurrence records and biome schemes, (2) choosing a
    biome scheme, (3) occurrences-to-biome classification, and (4)
    output and visualisation.