Nordic-Baltic Grassland Vegetation Database (NBGVD) – current state and future prospects

This Long Database Report describes the historical background and current contents of the Nordic-Baltic Grassland Vegetation Database (NBGVD) (GIVD-code EU-00-002). NBGVD is the EDGG-associated collaborative vegetation-plot database that collects vegetation-plot data of grasslands and other open


Introduction
Vegetation-plot databases have an enormous potential for vegetation ecology, macroecology and global-change studies (Dengler et al. 2011;Wiser 2016) as they allow for generalization beyond the local or regional extent.This potential is increasingly harvested through big continental to global databases such as the European Vegetation Archive (EVA; Chytrý et al. 2016), the global database "sPlot" (Bruelheide et al. 2019) or the specialised high-quality database of Palaearctic open habitats "GrassPlot" (Dengler et al. 2018).Publica-tions making use of data from EVA, for example, have shed light on the broad-scale classification of dry grassland types (Willner et al. 2019), the frequency and distribution of neophytes in European grasslands (Axmanová et al. 2021) and the relationship of functional vs. phylogenetic diversity in grasslands (Večeřa et al. 2023).The sPlot database, among others, was used to model how alpha-diversity patterns in grasslands and forests globally vary across grain sizes (Sabatini et al. 2022) and to test what drives stability in grasslands (Engel et al. 2023).However, such broad-scale analyses strongly depend on balanced data coverage across regions.

Scope of NBGVD
The Nordic and Baltic region in the sense of NBGVD is defined as the combined territories of Denmark, Faroe Islands, Iceland, Svalbard and Jan Mayen, Norway, Sweden, Finland, NW Russia, Belarus, Estonia, Latvia, Lithuania and the Pleistocene lowlands of N Poland and N Germany.This region approximately corresponds to the maximum extension of the Northern European ice shield during the Pleistocene (Lang 1994).However, German plots have recently been excluded from NBGVD and transferred to our sister database GrassVeg.DE (Dengler et al. 2017(Dengler et al. , 2018a)).In the future, we anticipate a similar arrangement with the Lithuanian Vegetation Database (EU-LT-001).
According to its Bylaws, NBGVD's main foci are "all natural and semi-natural grasslands s.l.".However, any vegetation types except forests, shrublands, true aquatic communities and arable fields are collected.According to Mucina et al. (2016), this means the following vegetation classes (although sources that contain a small fraction of other classes are digitised completely): • Lowland grassland and heathland communities: Calluno-Ulicetea, Festuco-Brometea, Koelerio-Coryne phoretea canescentis, Molinio-Arrhenatheretea, Nardetea, Sedo-Scleranthetea Further, the relevés must refer to contiguous plots with a specified area in the range of 0.09 to 400 m 2 .Relevés with a direct estimate of percent cover (see Dengler and Dembicz 2023) are preferred, but those with any other cover or cover-abundance measure (e.g.variants of the Braun-Blanquet or Hult-Sernander scales) are also accepted, while pure presence-absence data are not.

History and governance of NBGVD
The database originated from data collected by J. Dengler aimed at the phytosociological classification of the dry grasslands of the region, at that time managed in the software for vegetation plot handling SORT (Ackermann and Durka 1998).It was subsequently transferred to TUR-BOVEG 2.0 (Hennekens and Schaminée 2001) and be-came a collaborative project under the name "Database Dry Grasslands in the Nordic and Baltic Region" (see Dengler et al. 2006b;Dengler and Rūsiņa 2012).In 2016, the scope was widened to include all grasslands s.l., and thus, the name was changed to "Nordic-Baltic Grassland Vegetation Database" (NBGVD; Dengler and Kozub 2022).NBG-VD is registered in the Global Index of Vegetation-Plot Databases (GIVD; Dengler et al. 2011) under the ID EU-00-002 (see GIVD Fact Sheet).It is one of currently five regional grassland vegetation databases associated with the Eurasian Dry Grassland Group (EDGG; https://edgg.org/),namely the Balkan Dry Grassland Database (BDGD; EU-00-013; Vassilev et al. 2012), the German Grassland Vegetation Database (GrassVeg.DE; EU-DE-020; Dengler et al. 2017Dengler et al. , 2018)), the Romanian Grassland Database (RGD; EU-RO-008; Vassilev et al. 2018) and the Ukrainian Grassland Database (UGD; EU-UA-001; Kuzemko 2012).
NBGVD is a self-governed consortium in which every data contributor becomes a member.It is regulated by a set of Bylaws (Suppl.material 1).The members elect a Custodian and a Deputy Custodian for two-year renewable terms to coordinate the database, with J. Dengler the current Custodian and Ł. Kozub the Deputy.NBGVD contributes its content at regular intervals to EVA and sPlot, allowing its members to opt-in for EVA and sPlot publications as co-authors and propose studies using the continental and global databases themselves.The NBGVD website is https://edgg.org/databases/Regional-databases.
Starting with 7,675 plots in 2012 (Dengler and Rūsiņa 2012) and 9,839 plots in 2022 (Dengler and Kozub 2022), NBGVD has now grown to 12,694 plots as of March 2024.This means a net growth of 5,019 plots over12 years.However, the number of plots that were newly made available to EVA during this period was as high as 9,143, since more than 4,000 plots have since been excluded from NBGVD.Before 2022, all plots from Germany were transferred to GrassVeg.DE (Dengler et al. 2017(Dengler et al. , 2018)).Moreover, we recently excluded plots that are now included in a specialised EVA database from Latvia, the Semi-natural Grassland Vegetation Database of Latvia (EU-LV-001), to avoid duplicates in EVA.Apart from this increase in quantity, we also took great efforts to improve the quality and consistency of the plot data.This included identifying and correcting erroneous entries, filling in important header data fields (e.g.vegetation class), and improving coordinate precision (in the predecessor database under SORT, coordinates were only given with 0.1° precision).

Data management
Currently, the database is managed using the latest version of the TURBOVEG 2.0 software (Hennekens and Schaminée 2001).When we encountered cover-(abundance) scales not predefined in TURBOVEG, we added their definitions, i.e. the symbol, min, max and mean cover values of each cover class to the respective definition table of the program.The taxonomy of the database is based on the built-in "Europe" species list relying on the Flora Europaea (Tutin et al. 1964(Tutin et al. -1993) ) with some additions of vascular plant taxa that could not be easily attributed to any of the already existing taxonomic concepts as well as bryophytes and lichens (for these groups, TURBOVEG "Europe" does not have a clearly documented source).The header data in NBGVD consist of five groups: (1) standard TURBOVEG fields, (2) fields required by EVA (see Chytrý et al. 2016), (3) fields required by sPlot (see Bruelheide et al. 2019), (4) fields needed for NBGVD project management, and (5) structural and environmental variables provided with the data.The fields of the categories (1)-( 4) are filled completely, while the fields of category ( 5) are created and filled upon availability and carefully curated for consistency (e.g.identical units).

Content of NBGVD
The 12,694 vegetation plots currently included in NBG-VD originate from data published by consortium members (19.5%) and other authors (45.6%), while the rest are unpublished relevés from consortium members (34.9%).In total, the NBGVD currently contains data from 124 different sources (Suppl.material 2) contributed or digitised by 27 contributors, four of which contributed more than 10% each (Suppl.material 3: table S3.1).
NBGVD also contains various header data fields for structural and environmental variables.Apart from several environmental header data fields with free text, such as land use and soil texture class, there are currently 11 measured environmental variables referring to topography and soil characteristics (Table 2).Among these, elevation, aspect, inclination and pH (H 2 O) are available for at least one quarter of the plots (Table 2).

Resume and outlook
With the current NBGVD update, the data coverage of grasslands s.l. in the Nordic and Baltic regions has significantly improved in EVA and sPlot, thus facilitating regional, continental, and global analyses of non-forest vegetation.Aside from coming from an underrepresented region, the data in NBGVD are particularly useful for their good quality regarding the treatment of non-vascular plants (currently in more than 80% of all plots and thus high above the EVA average) and the careful curation of environmental header data.However, plot sizes could be a challenge when analysing data from the Nordic-Baltic region together with data from the rest of Europe, as the median plot sizes used in most of the NBGVD countries (4 m 2 or even 1 m 2 ) are well below the prevailing practice in most other European countries (Chytrý and Otýpková 2003).The data of NBGVD can be requested from NBGVD directly or via EVA and sPlot, following their respective Bylaws.In fact, they are frequently used in EVA and sPlot projects.However, in early 2024, we received the first direct data request to NBGVD, interestingly for a pan-European study where the proponent combined the European grassland data from sPlotOpen (a stratified, open access subset of an older EVA release: Sabatini et al. 2021) and GrassPlot (Dengler et al. 2018b), but wanted to achieve a data coverage in the Northern countries comparable to the rest of Europe.
Even with the release of this NBGVD update, the data coverage in the region is still poorer than in many other parts of Europe.Thus, we aim at expanding NBG-VD further through the inclusion of additional data which could stem from the digitisation of older published sources from the past or from recent data (published and unpublished) in digital format from current and new consortium members.Promising avenues to retrieve further historical sources are searching for the sources underlying the few synthetic vegetation overviews of the region (e.g.Dierßen and Dierßen 1996) and systematic screening of botanical and ecological journals of the region (e.g.Acta Phytogeographica Suecica, Blyttia,…).Anyone who has data to contribute that match our scope is welcome to contact J.D. or Ł.K.However, please note that NBGVD is an all-purpose database with minimal requirements for data to be provided (see above).If you have data that meet the higher standards of our partner database GrassPlot (Dengler et al. 2018;Biurrun et al. 2019), particularly if they have been sampled on precisely delimited plots of one of the GrassPlot standard sizes (e.g.0.1, 1, 10 or 100 m 2 ), we recommend contributing these to Grass-Plot, not to NBGVD.In this case they would not only be available for EVA and sPlot projects, but also for Grass-Plot projects (see https://edgg.org/databases/GrassPlot)(Figure 5).Likewise, data from repeatedly sampled plots (permanent or quasi-permanent plots) would likely be more beneficial if contributed to ReSurveyEurope Figure 5. Decision tree (grey) on how to contribute grassland plots s.l.from the Nordic-Baltic region to national and international vegetation-plot databases to achieve optimal benefit for data contributors and for science.The black lines indicate the flow of data to databases of higher aggregation levels.(Knollová et al. 2024) (Figure 5).Note that despite belonging to the Nordic-Baltic region, grassland plots s.l.from Germany not meeting the GrassPlot criteria should go to GrassVeg.DE, and those from Lithuania and the semi-natural grasslands from Latvia preferentially to the respective national databases (Figure 5).
Since NBGVD has no funding, we rely on voluntary work both for digitising and georeferencing plots and for further improvement of the quality of already included plots (e.g. to increase the georeferencing precision of plots provided in the past).We also would like to adjust the taxonomic backbone to the current European standards of vascular plants, bryophytes and lichens to allow the most effective use of the data.Here, the "Euro+Med augmented" standard from Dengler et al. (2023), based on Euro+Med (2023) for vascular plants and Hodgetts et al. (2020) for bryophytes, appears particularly promising.
We anticipate that having more high-quality plot data from the Nordic-Baltic region will not only improve the validity of future macroecological and global-change studies for this region, but be particularly beneficial for the development of data-based broadscale vegetation classification systems, of which there are only a few for the open habitats of the region so far, namely for the vegetation of fens (Peterka et al. 2017), coastal dunes (Marcenò et al. 2018), bogs (Jiroušek et al. 2022), and springs (Peterka et al. 2023) across Europe and for the rocky outcrop communities in the Nordic-Baltic region (Dengler and Löbel 2006;Dengler et al. 2006a).

Figure 1 .
Figure 1.Spatial distribution of the 12,694 vegetation plots contained in NBGVD in March 2024.The grey shading indicates the geographic scope of NBGVD, while the colour intensity of the 100 km × 100 km grid cells represents the number of plots contained in NBGVD.

Figure 2 .
Figure 2. Temporal distribution of the 12,694 vegetation plots contained in the NBGVD in March 2024.

Figure 3 .
Figure 3. Distribution of the plot sizes of the 12,694 vegetation plots contained in the NBGVD in March 2024.

Figure 4 .
Figure 4. Distribution of coordinate precision of the 12,694 vegetation plots contained in the NBGVD in March 2024.

Table 1 .
Countries and other territories covered by NBGVD with their area and available plot number in March 2024, expressed in absolute and relative terms and density per surface area.

Table 2 .
Measured environmental variables contained in NBGVD with their degree of availability and the distribution of values.