ReSurveyEurope: A database of resurveyed vegetation plots in Europe

Aims: We introduce ReSurveyEurope — a new data source of resurveyed vegetation plots in Europe, compiled by a collaborative network of vegetation scientists. We describe the scope of this initiative, provide an overview of currently available data, governance, data contribution rules, and accessibility. In addition, we outline further steps, including potential research questions. Results: ReSurveyEurope includes resurveyed vegetation plots from all habitats. Version 1.0 of ReSurveyEurope contains 283,135 observations


| INTRODUC TI ON
Plot-based vegetation samples, called vegetation plots henceforth, document plant species diversity and community composition at a given location at a given time.Such data allow for a wide variety of research questions to be addressed, and accordingly, they have been widely used in vegetation ecology for more than a century (e.g., Braun-Blanquet, 1964;Ellenberg & Leuschner, 2017).More recently, the mobilization of vegetation-plot data, and their compilation in large repositories such as the European Vegetation Archive (EVA, http:// eurov eg.org/ eva-database) and the global vegetation database sPlot (https:// www.idiv.de/ de/ splot.html) have substantially expanded the availability of these types of data (Schaminée et al., 2009;Dengler et al., 2011).
Vegetation plots also contain metadata (Mucina et al., 2000) that describe properties of the surveyed site (e.g., topography, land use, slope inclination and aspect), the plot (e.g., size, shape), and the sampling (e.g., date of sampling, surveyor name).In most cases, information on the location of the vegetation plot is provided, although this information may be given in a variety of ways and with different geographic accuracies, ranging from vague textual descriptions of localities to highly accurate geographic coordinates based on GPS data or permanent marks in the field.
Increasing human pressures on the natural environment during the last decades have caused substantial changes in the European flora and vegetation (e.g., Richner et al., 2015;Jandt, Bruelheide, Jansen, et al., 2022).Vegetation-plot data offer excellent opportunities to study temporal vegetation changes with fine resolution across large areas by resurveying historical vegetation plots, or by establishing permanent plots that are revisited on a regular basis.
The number of publications (and corresponding data sets) based on resurveyed vegetation plots in Europe (and elsewhere) has increased strongly since the start of the new millennium (Figure 1).
While some initiatives with a focus on specific habitats such as forests (forestREplot, https:// fores trepl ot.ugent.be/ ) and alpine environments (GLORIA, https:// www.gloria.ac.at/ hom/ ) have integrated a part of these data, many resurvey data sets are not yet integrated into large data repositories that ensure easy access and interoperability of the data.As a consequence, the potential of using existing data to address research questions remains limited.Therefore, we launched ReSurveyEurope in 2020 to fill this gap at the European level.ReSurveyEurope is closely connected to the European Vegetation Archive and aims to build a data repository of fine-scale plot-based plant community resurvey data (relevés, plots).Here, we describe the scope of this initiative, provide an overview on data, governance rules, accessibility, and outline further steps of how to ensure that ReSurveyEurope stays a living database.

| Scope of ReSurveyEurope
The geographic scope of ReSurveyEurope is the continent of Europe, including adjacent, biogeographically similar areas such as the Macaronesian Islands (excluding Cape Verde), Greenland, Georgia, Armenia, Azerbaijan, Anatolia, Cyprus, and the Mediterranean Biome in northern Africa and the Near East.Suitable data sets are fine-grain biodiversity data (e.g., vegetation plots, relevés, transects) from all habitats that document the presence (and ideally also proxies of the cover) of vascular plants (and optionally bryophytes and lichens) with at least two repeated observations (i.e., individual surveys of each plot) in different years.Observations have to be made with identical or at least similar and comparable methods at the same or a similar location close by.
To be suitable for ReSurveyEurope, data sets have to fulfill a set of minimum requirements: (i) a complete list of vascular plants in defined plots, preferably also with information on species cover or proxies for cover such as biomass; (ii) plot location including longitude and latitude, and information on the precision of the coordinates; (iii) sampling date (ideally the precise date, minimum the year); and (iv) plot size (see Appendix S1 for details).Further, information on the methods of data collection, total vegetation cover, elevation, slope inclination and aspect, bedrock, land use, recent land-use change and vegetation or habitat type are welcome to increase the value of the data set.

| Data preparation, integration and taxonomic standardization
The first data call for ReSurveyEurope was announced on October 6, 2020 (http:// eurov eg.org/ downl oad/ ReSur veyEu rope_ EVA_ Call.pdf).Data contributors were asked to ideally provide their data in Turboveg 2 format (Hennekens & Schaminée, 2001), with a set of predefined header data fields following the guidelines in Appendix S1.
For easier data preparation, providers can download the database structure with the standard database fields from https:// www.synbi osys.alter ra.nl/ turbo veg/ (under Database dictionaries/ReSurvey) using Turboveg 2 version 2.148 or higher.
All databases in ReSurveyEurope are stored and managed in Turboveg 2 using national species lists as the taxonomic backbone for the individual databases.The header data were checked and, if necessary, standardized by the first author of this article before integrating the data sets into ReSurveyEurope.As a next step, all databases were pooled in a single database in Turboveg 3, which has been developed for EVA by S.M. Hennekens to facilitate joint management of databases using different species lists and database dictionaries.

| Description and characterization of the plots and observations
In total, version 1.0 of ReSurveyEurope contains 79,190 plots with 283,135 observations (i.e., resurveys of individual plots) from 449 independent resurvey projects (Figure 2; see https:// eurov eg.org/ resur vey/ database for a full list).Of the total number of plots, 62,139 (78%) are from permanent (incl.manipulated) plots, for example, marked or geotagged plots which allow for high spatial accuracy in resurveying (Figure 2b,c).Note that this category also includes plots that are not permanently marked in the field, but were located with high precision (e.g., GPS-measured locations, photos).
The remaining 17,051 (22%) plots are quasi-permanent plots, which in the initial survey had not been sufficiently georeferenced to guarantee their unambiguous retrieval (Figure 2d).Resurveys of such plots rely on the geographic information available from the first survey including geographic coordinates, textual description of plot locations, information from field notes and maps, and surveyors' knowledge of plot locations.These plots are thus associated with higher uncertainties on the location of the initial plots.
Four data sets accounting for 28,470 (35%) plots provide only presence/absence information on plant species; one data set from Denmark is especially large and accounts for most of the presence/ absence-only data; however, for this data set, abundance information will be provided in the near future.Of the remaining 50,720 (65%) plots, the majority contain abundance information (e.g., percentage cover or cover-abundance classes such as variants of the Braun-Blanquet scale) with some plots from mixed data sets (i.e., containing abundance information for a subset of plots).
The geographical distribution of plots included in ReSurveyEurope is uneven, with central and northwestern European countries contributing the most data so far (Table 1; Figure 2; Appendix S2).For the full data set (including abundance and presence/absence data sets), Denmark ranks first by a wide margin (35%), followed by the United Kingdom (17%), Germany (15%), Switzerland (7%), and Poland (3%).When only considering the plots with species abundance information, the highest percentages are for the United Kingdom (28%), followed by Germany (25%), Switzerland (10%), Poland (5%), and Spain (4%).

F I G U R E 1
Cumulative number of published studies on vegetation change including studies using resurveyed vegetation plots (green) based on 1345 publications indexed in Web of Science (as of December 6, 2022) following the same search criteria as Vellend et al. (2013), compared to the overall number of publications within the field of ecology (blue).Of the former, 420 publications are from 1990-2012 (i.e., the period covered by Vellend et al., 2013; in light green) and an additional 925 papers were published thereafter (in dark green) - an increase by 220% during the last decade.In comparison, overall publications within the field of ecology have only increased by 146% (following a Web of Science search based on the search string: "plant" AND "species" AND "ecology").The search string used here and by Vellend et al. (2013) was: ("plant community" OR "plant communities" OR "vegetation" OR "forest*" OR "grassland*" OR "wetland*" OR "desert*" OR "savanna*" OR "tundra*" OR "steppe*" OR "shrubland*" OR "prairie*" OR "taiga" OR "rainforest*" OR "woodland*" OR "mangrove*") AND TITLE-ABS-KEY ("biodiversity" OR "diversity" OR "richness" OR "evenness" OR "composition") AND ALL ("resurvey*" OR "resample*" OR "revisit*" OR "temporal change").The assignment of the data in ReSurveyEurope 1.0 to EUNIS Level 1 habitat classes using the classification expert system EUNIS-ESy (Chytrý et al., 2020) (Figure 4) shows that for 36,333 plots (46%) an assignment to a EUNIS habitat class was not possible - most often this was the case because the assignment to a habitat type is not possible for plots without species cover information.A total of 21,357 plots (27%) were assigned to grasslands, followed by 8976 Plot size differed substantially between plots, with plots located in forests being larger on average than plots in non-forest vegetation types (Figure 5b).In non-forest vegetation, median plot size is 2 m 2 (minimum: 0.001 m 2 ; maximum: 2500 m 2 , mean: 26 m 2 ), while the median plot size for forest plots is 100 m 2 (minimum: 3 m 2 ; maximum: 2500 m 2 , mean: 153 m 2 ).Spatial uncertainty of plot locations varies between 0 and 15,000 m (Figure 5c).By far the highest number of observations has     The number of observations per plot varies among data sets.The majority of data sets have been surveyed twice (mostly resurveying studies) or at most a few times (mostly studies with permanent plots), while a few, mostly small data sets have high numbers of observations (Table 2).

| Taxonomic and nomenclatural standards and recommendations for data usage
ReSurveyEurope applied the EVA approach of simultaneously using a set of nationally used plant taxonomic and nomenclatural concepts to best represent individual data sets.A total of 31 species lists are used for storing species data in particular databases.
Some species lists are fully taxonomically standardized and follow taxonomic concepts of particular countries (e.g., GermanySL 1.The harmonized nomenclature for vascular plants according to Euro+Med can be further adjusted using the expert system EUNIS-ESy (Chytrý et al., 2020) to merge multiple taxonomic levels (e.g., subspecies) to the level of species and aggregates for groups of closely related species that are often identified inconsistently across vegetation plots.This adjustment can be performed using JUICE 7 (Tichý, 2002), Turboveg 3 (Hennekens, 2015) or an R script developed by Bruelheide et al. (2021).Before analyses, different observations from the same plot should also be checked for possible inconsistencies (e.g., the same taxon can be identified to the level of aggregate in the first sampling and to the level of species in the second sampling).In any case of doubt about correct taxonomic interpretation, the users should check the taxonomy and nomenclature used in the original data set.

| Cooperations and side projects
ReSurveyEurope makes use of pre-existing infrastructure, procedures and expertise developed and employed by EVA, and it is embedded in the EVA governance structure.Data integration and storage are coordinated and conducted at Masaryk University in Brno.EVA data are formally divided into EVA core databases (nonrepeated observations) and ReSurveyEurope.There is some overlap between these two subsets.Some databases contain mostly non-repeated observations but also a small part of repeatedly F I G U R E 4 Distribution of 42,837 plots in ReSurveyEurope across habitat types defined as EUNIS Level 1 habitat classes classified using EUNIS-ESy (Chytrý et al., 2020).The category "habitat change" refers to plots that were assigned to different habitat types over the course of their observations.The remaining plots (not shown in the figure) are those without species cover information, which cannot be classified using EUNIS-ESy.

| Strengths, gaps, limitations, and pitfalls
ReSurveyEurope represents the first European data set of vegetation plots resurveyed at least once from a wide range of habitats.
The original data were in most cases sampled for different purposes, by using a range of methodologies applied within the respective projects.Despite the harmonization of basic attributes upon integration of the data sets into ReSurveyEurope, there is substantial variation in many relevant features such as the accuracy of plot location, plot  Accordingly, it is crucial to clarify which data are suitable for specific analyses and to select - and possibly edit - data sets accordingly (Chytrý et al., 2014;Kapfer et al., 2017).Perhaps most importantly, it is necessary to know whether particular plots represent non-manipulated vegetation, or experimental plots after some manipulation (e.g., nutrient addition, experimental disturbance) or a successional series.If the aim is to study spontaneous vegetation change, it may be necessary to select only control plots from experiments.Even then, researchers have to be careful about selecting the suitable subset of data (e.g., in regularly mown areas, experimental manipulation can be abandonment with a mown plot as a control, whereas in abandoned areas, manipulation can be mowing with an abandoned plot as a control).

| PROPERT Y RI G HTS , DATA USAG E AND ACCE SS
ReSurveyEurope is devoted to an inclusive and transparent governance, collaboration, co-authorship, and data usage approach.

| FUTURE DE VELOPMENTS AND OUTLO O K
ReSurveyEurope is intended to be a living and dynamic resource, and thus, we expect that it will further grow and mature in the years to come.While a large number of existing resurvey data sets from Europe have already been included in ReSurveyEurope 1.0, we expect and hope that further data sets will become mobilized and contributed.In addition, several studies resurveying historic vegetation-plot data are underway or have been proposed.Thus, we expect that further data sets will be sampled and, hopefully, contributed to ReSurveyEurope.As regular data updates are foreseen, we highly encourage colleagues to provide additional data once they become available.Submission of data sets is possible at any time (see https:// eurov eg.org/ resur vey/ for further information).
In addition, we believe that there is substantial potential for de- The oldest plot included was sampled in 1911 in the Swiss Alps (data set "DISEQUALP",Braun-Blanquet, 1913;Rumpf et al., 2018), while the latest resurveys are from 2022 (Figure3, Appendix S3).Thus, the total time span covered by ReSurveyEurope is 111 years.The mean time span between sampling the first and the last observation is 26 years across all the 449 projects.Mean starting dates within data sets are more recent, as often first observations of plots within a data set have been done in different years (Appendix S3).Four projects started with the first plots until 1930, 21 additional projects by 1950, and another 84 projects by 1970.The longest mean time span covered by one data set is 104 years (data set "DISEQUALP").

F
I G U R E 2 Geographic distribution of the 79,190 plots contained in ReSurveyEurope version 1.0 (a).Plots grouped by different study designs corresponding to (b) permanent plots, (c) manipulated permanent plots, and (d) quasi-permanent plots, in which plots of the initial survey have not been precisely georeferenced.16541103, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13235by Universitat Bern, Wiley Online Library on [04/07/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Overview of the top 10 countries ranked by plots included in ReSurveyEurope.Countries are ranked by the number of plots within their borders.For the full list of countries, see Appendix S2. a For the Russian Federation, only the European territory is considered.

F
I G U R E 3 Time period covered by the 449 individual projects in ReSurveyEurope 1.0.Projects with permanent, manipulated permanent and quasi-permanent plots are shown in different colors.Note that manipulated permanent projects may also contain some unmanipulated (=control) plots.Black points indicate sampling times.
photos.On the opposite side of uncertainty, there are 975 plots with an uncertainty of 100-1000 m, and 117 plots with an uncertainty greater than 1000 m.Typically, these spatially most uncertain observations were from plots established several decades ago and (re-)located post-hoc based on historical materials like maps or descriptive textual information given in the original source.
other lists are not fully standardized.In different species lists, the same name can refer to different taxa, different names can refer to the same taxon, and taxa can be circumscribed and ranked differently.To unify different taxonomic concepts, species lists were matched to the taxonomic and nomenclatural backbone of the Euro+Med PlantBase (Euro+Med, 2006) using the SynBioSys Taxon Database developed for EVA.Some taxa (e.g., non-European aliens, aggregates, unresolved names from the original data sets) that are not included in Euro+Med were retained in the data set.Both the original and harmonized taxonomies are preserved so that users can decide which of them they prefer to use.For cryptogams (lichens, bryophytes), taxonomy and nomenclature are so far largely unresolved across data sets in EVA and ReSurveyEurope.
sampled plots included in ReSurveyEurope.Other databases (e.g., EU-GB-004 FloodplainMeadows, EU-GB-005 Scottish Coastal Survey or EU-GB-006 Scottish Vegetation Resurvey) include only repeated plots but from vegetation types and areas that are otherwise poorly represented in EVA, and therefore they are included both in EVA core databases and ReSurveyEurope.Both EVA core databases and ReSurveyEurope can be provided for projects using the same data request form, in which the project proponent indicates the data selection criteria.ReSurveyEurope has established a number of cooperation agreements on data-sharing and possible joint analyses with other initiatives.These include, in particular: forestREplot, a database of forest resurvey plots in temperate zones (Verheyen et al., 2017; https:// fores trepl ot.ugent.be/ ), the Global Observation Research Initiative in Alpine Environments (GLORIA) network (Pauli et al., 2015; https:// www.gloria.ac.at/ home), and the Long-term Vegetation Sampling (LOTVS; Sperandii et al., 2022; https:// lotvs.csic.es/ ).Most European data sets from these global networks were integrated into ReSurveyEurope with the help of database managers and based on the consent of owners or providers of the individual original data sets.ReSurveyEurope also integrated the ReSurveyGermany database, developed in parallel with the development of ReSurveyEurope and using the same structure (Jandt, Bruelheide, Berg, et al., 2022).Further, a cooperation has been initiated with the Nutrient Status Initiative, which explores the role of nutrient availability on local plant diversity.A joint data call has been launched (http:// eurov eg.org/ downl oad/ ReSur veyEu rope_ Nutri entSt atus_ DataC all20 21. pdf).

F
Key attributes of plots (n = 79,190) in ReSurveyEurope 1.0.(a) Percentage of resurveying methods, 115 plots with mixed methods of resurveying not shown (see text).(b) Plot size distribution (forests are given as a separate category as the plot sizes are usually larger).(c) Distribution of the plotlocation uncertainty.

Further
, individual studies should only compare changes over identical periods, in terms of both the interval between two samplings and between the starting or mean date.For example, two resurvey data sets, one collected between the 1960s and 1980s and the other collected between the 1990s and 2010s, can show different trends, although the interval length between the first and last survey is the same.Resurveyed vegetation plots provide a valuable resource for the study of vegetation dynamics and their responses to environmental change in Europe.A particular strength of ReSurveyEurope is that empirical time-series data are provided for largely exact locations, which makes them ideal for calibrating or validating models.For example, the plot records might be linked to remote-sensing information on land-use change or change in habitat quality.They may also allow assessing the change in the conservation status of Annex 1 habitats of the EU Habitats Directive.At times of global climate and land-use change, ReSurveyEurope is an important archive of past vegetation composition.Such data can be used to inform conservation and management decisions, improve our understanding of ecosystem dynamics and help develop strategies to mitigate the impacts of climate change.
veloping further collaborations or joint activities with colleagues and initiatives to expand the scope of ReSurveyEurope.This may include (i) expanding the geographic scope, (ii) expanding the taxonomic scope, and (iii) standardizing and expanding metadata content in ReSurveyEurope.To conclude, the ReSurveyEurope consortium is open and interested in discussing suggestions for extending the initiative and thus encourages colleagues to bring any such ideas forward.As ReSurveyEurope data have become ready for use, several projects have already been initiated, and further proposals for analyses can be submitted to the coordinators.AUTH O R CO NTR I B UTI O N S Franz Essl, Ilona Knollová, Milan Chytrý, Helge Bruelheide, and Stefan Dullinger conceived the original idea for ReSurveyEurope, which was jointly developed by a core team also including Bernd Lenzner, Ilona Knollová, Florian Jansen, Ute Jandt and Stephan Hennekens.Stephan Hennekens, Milan Chytrý, Helge Bruelheide, Ute Jandt and Ilona Knollová designed the database structure.Ilona Knollová compiled the database with help from Stephan Hennekens.Ilona Knollová, Ekin Kaplan, Michael Glaser and Bernd Lenzner created the figures and tables for this article.Franz Essl led the writing of the manuscript with substantial contributions of Milan Chytrý, Helge Bruelheide, Stefan Dullinger, Ute Jandt, and Ilona Knollová.All other authors (ordered alphabetically) contributed plot data and commented on an advanced version of the manuscript.All authors agreed with the final manuscript.and Zoology, Faculty of Science, Masaryk University, Brno, Czech Republic 2 Institute of Biology/Geobotany and Botanical Garden, Martin Luther 16541103, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13235by Universitat Bern, Wiley Online Library on [04/07/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 16541103, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13235by Universitat Bern, Wiley Online Library on [04/07/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

| 11 of 18 Journal of Vegetation Science
16541103, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13235by Universitat Bern, Wiley Online Library on [04/07/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Overview on the 10 largest projects by number of plots included in ReSurveyEurope 1.0.The full list of projects is given in Appendix S3. 16541103, 2024, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13235by Universitat Bern, Wiley Online Library on [04/07/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License TA B L E 2