GrassBase, a morphology and life history dataset for Poaceae

# GrassBase overview The GrassBase dataset contains data on 1,170 morphological and life history traits for 11,369 species of Poaceae (grasses), matching 11,035 accepted species (at the time of this data release in 2026). The original database was compiled from c. 1986 to 2016 by W. Derek Clayton and colleagues at the Royal Botanic Gardens, Kew, using the Descriptive Language for Taxonomy (DELTA). GrassBase was intended for taxonomic research use and as an identification aid, but the data have also been used in comparative analyses. We here provide the original DELTA files for GrassBase as well as HTML species and genus descriptions as previously published on the Royal Botanic Gardens, Kew, website. We also provide tabular data files with extracted data along with the R scripts used for data extraction. # License, references, authors The GrassBase data are released under the Creative Commons Attribution 4.0 International license: re-use allowed provided the source is cited. Please cite the Zenodo dataset and the accompanying data paper if you use the data in your publication: - [Zotero citation] - [paper citation] Further GrassBase publications: - Harman K.T., Clayton W.D. 2007. Recent Developments in Kew’s Grass Databases (Grassbase). Kew Bulletin 62: 375–379. https://www.jstor.org/stable/20443365 - Renvoize S., Clayton D., Cope T., Williamson H. 2007. An electronic world grass flora. Aliso: A Journal of Systematic and Floristic Botany 23: 267–270. https://doi.org/10.5642/aliso.20072301.21 - Vorontsova M.S., Clayton D., Simon B.K. 2015. Grassroots e-floras in the Poaceae: growing GrassBase and GrassWorld. PhytoKeys 48: 73–84. https://doi.org/10.3897/phytokeys.48.7159 Original GrassBase authors: W. Derek Clayton, Kehan Harman, David Simpson, Maria S. Vorontsova (all Royal Botanic Gardens, Kew). Standards for the DELTA format are available via: https://www.tdwg.org/standards/delta/ # Original sources The trait data in GrassBase were originally sourced from Floras and other taxonomic literature. The DELTA file contains a character "source and year of coding or revision" (`sourceAndYearOfCodingOrRevision` in data table`grassbase_complete.tsv`) with short source descriptions (e.g. "`2nd Jap. Bam. 2001`") but no detailed bibliography was maintained in GrassBase. # Usage notes We provide the GrassBase data in three versions. - The *DELTA files* contain the original dataset entries in the [Description Language for Taxonomy (DELTA)](https://www.tdwg.org/standards/delta/). These are text files but require custom scripts or programs to be parsed. - The *HTML description pages* contain the data translated into taxonomic descriptions for both species and genera. These were previously available via the Royal Botanic Gardens, Kew, website. They can be displayed and navigated in a web browser. - The *tabular data files* contain the data extracted to tables that can be opened using spreadsheet software, or imported into environments like R for data analysis. Note that 334 currently accepted species names match more than one synonym (rows) in the original GrassBase. It is up to the user to decide whether/how to combine data in these cases. We provide currently accepted names in the file `grassbase_species.tsv`. # Files ## DELTA files - `grassbase_delta_files.zip`: Compressed archive containing the original GrassBase data in DELTA format. It consists of four files: - `bigitems`: the coded character data per species. - `chars`: definitions of characters and states. - `specs`: metadata, including character dependencies. - `grassbase_species_names.tsv`: tab-separated file containing names, authors and synonyms for Poaceae species, maintained as part of GrassBase. Note this reflects taxonomy as of the last GrassBase update in 2016 and is now outdated. Refer to file `grassbase_species.tsv` (see below) for updated names and World Flora Online identifiers. ## HTML files - `grassbase_html_descriptions.zip`: Compressed archive containing GrassBase species and genus descriptions as HTML files that can be navigated in a web browser. ## Tabular files: All files are tab-separated text files in UTF-8 encoding. Cells containing text characters are enclosed in double-quotes. Missing data are coded as `NA`. Comments are enclosed in less-than and greater-than brackets (`<>`). - `grassbase_characters.tsv`: Table containing metadata about the 1,090 characters in GrassBase, of which 1,070 are morphological and life history traits and the last 20 distribution, taxonomy and metadata. Columns: - `characterNo`: character number. - `characterName`: short character name used in data files. - `characterOriginalDescription`: original, complete description (including units for numeric characters). - `characterType`: "discrete-ordered", "discrete-unordered", "integer" or "numeric". - `isMandatary`: "yes" if character had to be coded in the original GrassBase (note that these characters are nevertheless not complete in some cases). - `hasImplicitState`: "yes" if character has an implicit (default) state assumed when no information was actively entered. - `inapplicableIf`: other character states (character number,state number) that render this character inapplicable, separated by colons. - `category`: grouping category for characters (not originally part of GrassBase). - `grassbase_species.tsv`: Table containing metadata about the 11,369 species with data available in GrassBase. Columns: - `speciesGrassBase`: original species name used in GrassBase. - `scientificName`: currently accepted species name. - `scientificNameAuthorship`: nomenclatural authority for currently accepted name. - `worldFloraOnlineID`: World Flora Online identifier. - `genus`, `tribe`, `subfamily`: classification according to the most recent Poaceae-wide classification (Soreng et al. 2022, Journal of Systematics and Evolution 60: 476-521, https://doi.org/10.1111/jse.12847). - `grassbase_states.tsv`: Table containing a total of 3,726 states for 861 discrete traits. Columns: - `characterNo`: character number. - `characterName`: character name. - `stateNo`: state number within character. - `stateDescription`: original description of state. Implicit states contain a comment `<<implicit>>`. - `grassbase_complete.tsv`: Table containing trait data for 11,313 species (rows, with species as row names) and all 1,090 characters (columns). Implicit state values were added and labeled with a comment `<i>`. Numerics and integers are textual descriptions, often with ranges and extreme values, e.g. `(1-) 3-8 (-10)`. Some discrete characters can take on multiple states, combined with `&`. - `grassbase_simplified.tsv`: Table containing trait data for 11,313 species (rows, with species as row names) and 1,070 characters (columns). Simplified version for comparative analysis. Only the morphology and life history columns 1-1,170 were retained. Numerics and integers represent single values, or means of ranges, when these were given; extreme values were ignored. All comments, including those identifying implicit values, were removed. - `grassbase_range_lower.tsv`, `grassbase_range_upper.tsv`, `grassbase_extreme_lower.tsv`, `grassbase_extreme_upper.tsv`: Extracted lower and upper ranges, and lower and upper extreme values for numeric and integer characters. Only rows and columns with data were retained. ## Scripts - `delta_extraction_scripts.zip`: Compressed archive of R scripts used for extracting the GrassBase data from DELTA files and formatting them as tabular files.