The basis of the survey scheme was a map of grid cells 1.3 × 1.01 km covering the Netherlands. Grid cells were assigned to participants by the project organisation. Each grid cell was then surveyed for several hours to one day aiming to make a complete list of all wild vascular plants occurring in the area. Survey data were recorded on field forms with abbreviations of scientific taxon names printed on them (Fig. 1). Nomenclature followed the second edition of the Prodromus Florae Batavae (Vuyck 1901). Additionally, miscellaneous observations, vegetation relevées and literature records (from 1832 until 1953) were submitted on special forms or in handwritten or typewritten letters. Most observations include the grid cell code, taxon name, date and up to 9 names of co-observers (Table 1). In total, 56,103 forms were digitised, of which 47,060 were field survey forms, 8,279 written notes and 764 vegetation relevées. The average number of taxa per form was 47. Most of the field forms contained higher numbers of taxa, while written notes usually reported only 1–5 taxa (Table 2).

Figure 1. 

Samples of the FLORIVON survey forms: A field form showing a square with a drawing of the surveyed area, space for writing down grid square code (‘hokje’), location name, observer name(s) and date, followed by two pages of taxon abbreviations that surveyors had to cross out after observation B written note with header data containing the grid cell code (e.g. L6.12.31), observer’s name and the survey date. Stamps confirm that the data have been included in printed atlas volumes.

Survey forms were digitised using Turboveg (Hennekens and Schaminée 2001), a computer programme usually used for handling phytosociological relevées, with customised species dictionaries matching the taxonomy and nomenclature of the field survey forms. All additional written information on the forms, including additional taxon names of, for example, non-native taxa, additional survey dates and remarks were temporarily included in the Turboveg header record and extracted afterwards.

Taxon names were mapped to current names using a translation table between the Prodromus Flora Batavae (Vuyck 1901) and a more recent checklist of vascular plants in the Netherlands (Groen et al. 1999). The original taxon name or its abbreviation is kept in the database. Grid cells codes were translated to geographical coordinates. Observer names were mapped to existing observer identifiers in the National Database Flora and Fauna.

Records without an observation date were assigned to the entire survey period of 1902–1950. Records without a valid taxon name or missing grid cell codes were omitted from the final dataset. Records with locations entirely outside the country or in the sea were also omitted. A total of 5,530 records were cleaned. The number of digitised observations after this first data cleaning step was 2,638,919.

Validation of the digitised observations was performed with an automated procedure which involved trying to find a match for each observation in a dataset, based on printed volumes of the Atlas of the Flora of the Netherlands (Mennema et al. 1980) and other digitised literature and collection records in the National Database Flora and Fauna, which had been validated in the past.

In the FLORIVON dataset, 142,838 observations did not match validated data sources and were considered for a manual review. Of the remaining unmatched observations, 110,889 records of common taxa were validated, i.e. taxa occurring in 30% or more of the 5 × 5 km grid squares in the Netherlands. A total of 2,415 records of less common taxa were validated if they were present in neighbouring grid cells. Further unmatched records, rare taxa, were validated by Gerard Dirkse by plotting them on a map for visual interpretation (17,427 observations). These observations were validated if they matched the geographical pattern of all other valid observations of the taxon. Herbarium specimens and publications mentioning an observation were also taken into account during validation. In the validation process, 12,107 out of 142,838 records were deleted (ca. 3%).

The validated dataset was added to the NDFF Verspreidingsatlas (http://www.verspreidingsatlas.nl), which is the platform FLORON uses to curate datasets. Simultaneously, the dataset was published through the GBIF Integrated Publishing Toolkit (IPT).

Joop van Heeswijk performed the digitisation between 2001 and 2018 as voluntary work. Laurens Sparrius performed the validation of the dataset. Gerard Dirkse assisted with the validation of non-native and doubtful taxa. Naturalis Biodiversity Center (Leiden) is hosting the physical archive with field forms and notes.

The dataset is curated on the NDFF Verspreidingsatlas data platform and will be updated on GBIF annually if any changes are made. Included Darwin Core terms are: occurrenceID, type, language, licenserightsHolder, accessRights, references, datasetName, basisOfRecord, eventDate, decimalLatitude, decimalLongitude, geodeticDatum, coordinateUncertaintyInMeters, scientificName, kingdom, taxonRank, scientificNameAuthorship.

Excluded information: Complete observer biographies, source type (field list, publication, specimen, vegetation relevée), location names and remarks were not included in the published dataset, but can be found in the source (curation) database, which can be accessed with the link below. This information was excluded due to privacy reasons or because it was deemed irrelevant.

Object name: FLORIVON

Format name: Darwin Core Archive format

Format version: 1.0

Character encoding: UTF-8

Language: English

Licence: http://creativecommons.org/licenses/by-nc/4.0/legalcode

First publication date: 2019/09/01

Distribution: http://www.verspreidingsatlas.nl:8080/ipt

DOI: https://doi.org/10.15468/ke2ody

Curation website: https://www.verspreidingsatlas.nl/waarnemingen

Number of records: 2,626,773

The dataset only includes taxa of vascular plants (Kingdom Plantae: clade Tracheophyta). Most of the taxa are native to the Netherlands. Occasionally, non-native taxa were recorded. Nomenclature follows the last edition of the Flora of the Netherlands (van der Meijden 2005). Non-native taxa not listed in this Flora follow The Plant List (The Plant List 2013).

The dataset contains distribution data of 2502 taxa at species or intraspecific level divided over 138 plant families. The plant families with the most observations in the dataset belong to the Asteraceae and Poaceae (Table 3).

Some taxa in FLORIVON are currently accepted as lumped taxa, which makes it impossible to compare taxon distributions for certain taxa (Table 4).

The dataset contains observations and literature data from 1832 to 1953. Most of the data were collected between 1902 and 1950 as part of the FLORIVON citizen science project (Fig. 2).

Figure 2. 

Number of surveys per 5 year period during the course of FLORIVON project.

The dataset covers the entire country of the Netherlands as it was in the period 1902–1950. At that time, the southern part of the province of Flevoland did not yet exist (Hoeksema 2007). Additionally, minor changes were made to the border with Germany and Belgium after World War II (Wijchgel 2008).

Survey data were collected in small grid cells of 1.3 × 1.01 km (kwartierhok) (Fig. 3), 16 of which can be combined into a larger grid cell of 5.0 × 4.167 km (uurhok), which is used on some forms. The grid system was created in 1902 by the botanical community itself because, until 1920, a km grid was lacking on the topographical military maps. These grid cells differ from the currently used grid, in which the smallest grid cells are 1 × 1 km and follow the Dutch National Coordinate Reference System (ESPG: 28992).

The periods before and after 1925 show different patterns of survey intensity, which should be taken into account when using the data for further analysis (Fig. 4).

Figure 3. 

Example of a FLORIVON grid cell of 1.3 × 1.01 km, the smallest spatial unit in which data were collected. Map: OpenStreetMap.

Figure 4. 

Maps of the Netherlands showing the number of taxa recorded per grid cell before and after 1925 in the context of the FLORIVON project.