Where are we now with European forest multi-taxon biodiversity and where can we head to?

Biological Conservation 284 (2023) 110176 Available online 30 June 2023 0006-3207/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/). Review Where are we now with European forest multi-taxon biodiversity and where can we head to? Sabina Burrascano a , b , * , Francesco Chianucci b , c , Giovanni Trentanovi d , Sebastian Kepfer-Rojas e , Tommaso Sitzia b , f , Fl´ ora Tinya g , Inken Doerfler h , i , Yoan Paillet j , Thomas Andrew Nagel k , Bozena Mitic l , Lourdes Morillas m , n , Silvana Munzi n , p , Theo Van der Sluis q , Edoardo Alterio f , Lorenzo Balducci a , Rafael Barreto de Andrade s , Christophe Bouget t , Paolo Giordani u , Thibault Lachat v , w , Dinka Matosevic x , Francesca Napoleone a , Juri Nascimbene y , Chiara Paniccia z , Nicolas Roth v , w , R´ eka Aszal´ os g , Gediminas Brazaitis ab , Andrea Cutini c , Ettore D’Andrea b , ac , Pallieter De Smedt ad , Jacob Heilmann-Clausen ae , Philippe Janssen j , Daniel Koz´ ak af , Anders Mårell ag , Martin Mikol´ aˇ s af , Bj¨ orn Nord´ en ah , Radim Matula af , Peter Schall ai , Miroslav Svoboda af , Mariana Ujhazyova aj , Kris Vandekerkhove ak , Michael Wohlwend al , Fotios Xystrakis am , Michele Aleffi an , Christian Ammer ai , Frederic Archaux ag , Thomas Asbeck ao , Dimitrios Avtzis ap , Manfred Ayasse aq , Simonetta Bagella ar , Rosario Balestrieri as , Anna Barbati at , Marco Basile au , Ariel Bergamini au , Giada Bertini c , Alessandro Bruno Biscaccianti ax , Steffen Boch ay , J´ anos B¨ ol¨ oni g , Pierluigi Bombi az , Yves Boscardin ag , Giorgio Brunialti ba , Hans Henrik Bruun bb , François Buscot bc , bd , David Bille Byriel be , Thomas Campagnaro b , f , Alessandro Campanaro b , bf , Matthieu Chauvat bg , Michał Ciach bh , Marek ˇ Ciliak aj , Luca Cistrone bi , Joao Manuel Cordeiro Pereira bj , Rolf Daniel bk , Bruno De Cinti az , Gabriele De Filippo bl , Wouter Dekoninck bm , Umberto Di Salvatore bn , Yann Dumas ag , Zolt´ an Elek bo , cz , Fabrizio Ferretti c , Dimitrios Fotakis ap , Tam´ as Frank g , Julian Frey bp , Carmen Giancola bq , Erika Gomoryov´ a br , Marion Gosselin ag , Frederic Gosselin ag , Martin M. Gossner au , Frank G¨ otmark bs , Elena Haeler bt , Aslak Kappel Hansen bu , Lionel Hertzog bv , Jeˇ nýk Hofmeister bw , Jan Hoˇ sek bx , Vivian Kvist Johannsen be , Mathias Just Justensen be , Nathalie Korboulewsky ag , Bence Kov´ acs g , Ferenc Lakatos ca , Carlos Miguel Landivar cb , Luc Lens cc , Emanuele Lingua f , Fabio Lombardi cd , Frantiˇ sek M´ aliˇ s br , Luca Marchino c , Vitas Marozas cf , Giorgio Matteucci cg , Walter Mattioli ch , Peter Friis Møller ci , J¨ org Müller cj , ck , Csaba N´ emeth g , G´ abor ´ Onodi cl , Francesco Parisi b , bq , Thomas Perot ag , Sandrine Perret ag , Anna Maria Persiani a , Alessia Portaccio f , Mario Posillico cn , ˇ Zydr¯ unas Preikˇ sa cf , Carsten Rahbek ae , Nolan James Rappa bj , Sonia Ravera cp , Antonio Romano cq , Ferenc Samu cr , Christoph Scheidegger au , Inger Kappel Schmidt be , Sebastian Schwegmann bj , Flavia Sicuriello az , Andreea Petronela Spinu ct , Gavriil Spyroglou ap , Jonas Stillhard au , Eleni Topalidou ap , Anders P. Tøttrup cu , Karol Ujh´ azy cv , Katalin Veres g , Kris Verheyen ad , Wolfgang W. Weisser cw , Livia Zapponi cx , P´ eter ´ Odor g , cy a Sapienza University of Rome, 00185, Rome, Italy b National Biodiversity Future Center, Palermo, Italy * Corresponding author at: Sapienza University of Rome, 00185 Rome, Italy. E-mail address: [email protected] (S. Burrascano). Contents lists available at ScienceDirect Biological Conservation journal homepage: www.elsevier.com/locate/biocon https://doi.org/10.1016/j.biocon.2023.110176 Received 25 February 2023; Received in revised form 9 May 2023; Accepted 20 June 2023 Biological Conservation 284 (2023) 110176 2 c Council for Agricultural Research and Economics (CREA), Research Centre for Forestry and Wood, V.le Santa Margherita 80, I-52100 Arezzo (AR), Italy d Research Institute on Terrestrial Ecosystems, National Research Council, 50019 Sesto Fiorentino, FI, Italy e University of Copenhagen, Department of Geosciences and Natural Resource Management, 1958 Frederiksberg, Denmark f Department Land, Environment, Agriculture and Forestry, Universit` a degli Studi di Padova, 35020 Legnaro, PD, Italy g Centre for Ecological Research, Institute of Ecology and Botany, 2163 V´ acr´ at´ ot, Hungary h University of Oldenburg, 26129 Oldenburg, Germany i Technical University of Munich, Freising, Germany j Univ. Grenoble Alpes, INRAE, Lessem, 38402 Saint Martin d’Heres, France k University of Ljubljana, Department of Forestry and Renewable Forest Resources, Biotechnical Faculty, 1000 Ljubljana, Slovenia l University of Zagreb, Faculty of Science, Department of Biology, 10000 Zagreb, Croatia m University of Seville, Department of Vegetal biology and Ecology, 1749-016 Seville, Spain n Universidade de Lisboa, Center for Ecology, Evolution and Environmental Changes & CHANGE - Global Change and Sustainability Institute, Lisbon, Portugal p Centro Interuniversit´ ario de Hist´ oria das Ciˆ encias e da Tecnologia, Lisbon, Portugal q Wageningen Environmental Research, 6800 AA Wageningen, Netherlands s University of Maryland, 20742 College Park, United States t INRAE, UR EFNO, 45290 Nogent-sur-Vernisson, France u DIFAR, University of Genova, 16148 Genova, Italy v Bern University of Applied Sciences, School of Agricultural, Forest and Food Sciences HAFL, 3052 Zollikofen, Switzerland w Swiss Federal Research Institute WSL, Birmensdorf, Switzerland x Croatian Forest Research Institute, 10450 Jastrebarsko, Croatia y BIOME Lab, Department of Biological, Geological and Environmental Sciences, Alma Mater Studiorum, University of Bologna, 40126 Bologna, Italy z Institute for Alpine Environment, Eurac Research, 39100 Bolzano, Italy ab Department of Forest Science, Ariculture Academy, Vytautas Magnus University, 53361 Kaunas dist., Lithuania ac Research Institute on Terrestrial Ecosystems, National Research Council, 05010 Porano, Italy ad Ghent University, 9090 Melle-Gontrode, Belgium ae Center for Macroecology, Evolution and Climate, Globe Institute, University of Copenhagen, DK-2100 Copenhagen, Denmark af Czech University of life Sciences in Prague, 16500 Prague, Czechia ag INRAE, UR EFNO, 45290 Nogent-sur-Vernisson, France ah Norwegian Institute for Nature Research, 0855 Oslo, Norway ai Silviculture and Forest Ecology of the Temperate Zones, University of G¨ ottingen, 37077 G¨ ottingen, Germany aj Faculty of Ecology and Environmental Sciences, Technical University in Zvolen, 96001 Zvolen, Slovakia ak INBO-Research Institute for Nature and Forests, 9500 Geraardsbergen, Belgium al Albert-Ludwigs-University Freiburg, 79085 Freiburg, Germany am Forest Research Institute, Hellenic Agricultural Organization Demeter, 5700 Vassilika, Greece an School of Biosciences, University of Camerino, 62032 Camerino, MC, Italy ao Unique landuse GmbH, 79098 Freiburg, Germany ap Forest Research Institute, Hellenic Agricultural Organization Demeter, 57006 Vassilika, Greece aq Ulm University, Institute of Evolutionary Ecology and Conservation Genomics, Ulm, Germany ar University of Sassari, Department of Chemical, Physical, Mathematical and Natural Sciences, Sassari, Italy as Department of Marine Animal Conservation and Public Engagement, Anton Dohrn Zoological Station, 00015 Napoli, Italy at University of Tuscia, Department for Innovation in Biological, Agro-food and Forest systems (DIBAF), 01100 Viterbo, Italy au Swiss Federal Institute for Forest, Snow and Landscape Research WSL, 8903 Birmensdorf, Switzerland ax Dept. PAU, Mediterranea University of Reggio Calabria, 89124 Reggio Calabria, Italy ay WSL Swiss Federal Research Institute, 8903 Birmensdorf, Switzerland az Institute of Research on Terrestrial Ecosystems, National Research Council, Montelibretti, Italy ba TerraData Environmetrics, Spin-Off Company of the University of Siena, 58025 Monterotondo Marittimo, Italy bb University of Copenhagen, Dept Biology, 2100 Copenhagen, Denmark bc Helmholtz Centre for Environmental Research (UFZ), 06120 Halle (Saale), Germany bd German Center for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Leipzig, Germany be University of Copenhagen, Department of Geosciences and Natural Resource Management, Rolighedsvej 23, 1958 Frederiksberg C, Denmark bf Council for Agricultural Research and Economics (CREA), Research Centre for Plant Protection and Certification, Via Lanciola 12A, I-50125 Florence, Italy bg Normandie UNIV, UNIROUEN, INRAE, ECODIV, 76821 Mont Saint Aignan, France bh University of Agriculture in Krakow, Department of Forest Biodiversity, Krakow, Poland bi Forestry and Conservation, 03043 Cassino, Italy bj Albert-Ludwigs-University of Freiburg, Chair of Nature Conservation and Landscape Ecology, 79106 Freiburg, Germany bk Genomic and Applied Microbiology & G¨ ottingen Genomics Laboratory, Institute of Microbiology and Genetics, Georg-August University of G¨ ottingen, Grisebachstr. 8, 37077 G¨ ottingen, Germany bl Istitute of Wildlife Management, 80126 Napoli, Italy bm Royal Belgian Institute for Natural Sciences, Brussels, Belgium bn Council for Agricultural Research and Economics (CREA), Research Centre for Agricultural Policies and Bioeconomy, Via Lombardia, c.da Bucceri, I-65012 Cepagatti (PE), Italy bo Department of Biostatistics, University of Veterinary Medicine, Istv´ an utca 2, 1078, Budapest, Hungary bp Albert-Ludwigs-University of Freiburg, Chair of Forest Growth and Dendroecology, 79106 Freiburg, Germany bq Department of Biosciences and Territory, Universit` a degli Studi del Molise, Campobasso, Italy br Technical University in Zvolen, 96001 Zvolen, Slovakia bs University of Gothenburg, 40530 G¨ oteborg, Sweden bt Federal Research and Training Centre for Forests, Natural Hazards and Landscape (BFW), 1131 Vienna, Austria bu Natural History Museum of Denmark, Univeristy of Copenhagen, 2100 Copenhagen, Denmark bv Thünen Institute of Biodiversity, 38116 Braunschweig, Germany bw Czech University of Life Sciences Prague, Faculty of Forestry and Wood Sciences, Department of Forest Ecology, 16500 Praha, Czech Republic bx Ecological Services, 26801 Hoˇ rovice, Czech Republic ca University of Sopron, H-9400 Sopron, Hungary cb Albert-Ludwigs-University of Freiburg, Dept of Biometry and Environmental System Analysis, 79106 Freiburg, Germany cc Ghent University, Ghent, Belgium cd Agraria Department, Mediterranean University of Reggio Calabria, 89122 Reggio Calabria, Italy cf Agriculture Academy, Vytautas Magnus University, 53361 Kaunas distr., Lithuania cg National Research Council of Italy, Institute of BioEconomy (CNR-IBE), 50019 Sesto Fiorentino, FI, Italy ch Council for Agricultural Research and Economics (CREA), Research Centre for Forestry and Wood, Via Valle della Quistione 27, I-00166 Rome, Italy ci Geological Survey of Denmark and Greenland, 1350 Copenhagen K, Denmark cj University Würzburg, 96181 Rauhenebrach, Germany S. Burrascano et al. Biological Conservation 284 (2023) 110176 3 ck Nationalpark Bavarian Forest, Grafenau, Germany cl National Laboratory for Water Science and Water Security, Balaton Limnological Research Institute, H-8237 Tihany, Hungary cn Reparto Carabinieri Biodiversit` a Castel di Sangro, 67031 Castel di Sangro, Italy cp Department of Biological, Chemical and Pharmaceutical Sciences and Technologies (STEBICEF), University of Palermo, 90123 Palermo, Italy cq Institute for Bioeconomy, National Research Council of Italy, 00100 Roma, Italy cr Plant Protection Institute, Centre for Agricultural Research, ELKH, H-1029 Budapest, Hungary ct Albert-Ludwigs-University of Freiburg, Chair of Silviculture, 79106 Freiburg, Germany cu Natural History Museum of Denmark, Univeristy of Copenhagen, DK-1350 Copenhagen, Denmark cv Faculty of Forestry, Technical University in Zvolen, 96001 Zvolen, Slovakia cw Technical University of Munich, Terrestrial Ecology Research Group, Department of Life Science Systems, School of Life Sciences, 85354 Freising, Germany cx Institute of BioEconomy, National Research Council, 38098 San Michele All’adige, TN, Italy cy University of Sopron, Institute of Environmental Protection and Nature Conservation, Sopron, Hungary cz ELKH-DE Anthropocene Ecology Research Group, University of Debrecen, Egyetem t´ er 1, 4032, Debrecen, Hungary ARTICLE INFO Keywords: Forest biodiversity Multi-taxon Sustainable management Biodiversity conservation Forest stand structure ABSTRACT The European biodiversity and forest strategies rely on forest sustainable management (SFM) to conserve forest biodiversity. However, current sustainability assessments hardly account for direct biodiversity indicators. We focused on forest multi-taxon biodiversity to: i) gather and map the existing information; ii) identify knowledge and research gaps; iii) discuss its research potential. We established a research network to fit data on species, standing trees, lying deadwood and sampling unit description from 34 local datasets across 3591 sampling units. A total of 8724 species were represented, with the share of common and rare species varying across taxonomic classes: some included many species with several rare ones (e.g., Insecta); others (e.g., Bryopsida) were represented by few common species. Tree-related structural attributes were sampled in a subset of sampling units (2889; 2356; 2309 and 1388 respectively for diameter, height, deadwood and microhabitats). Overall, multitaxon studies are biased towards mature forests and may underrepresent the species related to other developmental phases. European forest compositional categories were all represented, but beech forests were overrepresented as compared to thermophilous and boreal forests. Most sampling units (94%) were referred to a habitat type of conservation concern. Existing information may support European conservation and SFM strategies in: (i) methodological harmonization and coordinated monitoring; (ii) definition and testing of SFM indicators and thresholds; (iii) data-driven assessment of the effects of environmental and management drivers on multi-taxon forest biological and functional diversity, (iv) multi-scale forest monitoring integrating in-situ and remotely sensed information. 1. Introduction Forests support about three-quarters of terrestrial plants, fungi and animal species (FAO, 2020), and are at the base of other ecosystem services including the provisioning of raw materials and the regulation of geochemical cycles. These services are threatened by climate change, forest loss and degradation, invasions by non-native species, and overharvesting (Felipe-Lucia et al., 2020). The increasing concern related to these threats has imposed a paradigm shift from single-objective forest management (i.e., timber production-oriented) to the embracement of forest multifunctionality (Mori et al., 2017). Sustainable Forest Management (SFM) is defined as the management that concomitantly maintains forest biodiversity, productivity, regeneration capacity, and vitality, as well as forests’ potential to fulfill a wide range of functions and services (MCPFE, 1993). As such, SFM is globally recognized as a crucial tool to counteract biodiversity loss, and to promote sustainable development (UN, 2015). Managing forests sustainably is particularly relevant in Europe, where, although about 24 % of forests are formally protected, only a small fraction (2 % of total forest area) is not subject to harvesting interventions (Forest Europe, 2020). Accordingly, the SFM definition reported in the European Union regulation (2020/852) includes a criterion of biodiversity maintenance, reported as the contribution to “enhancing biodiversity, halting or preventing the degradation of ecosystems, deforestation and habitat loss”. Nevertheless, the effects of forest management on the diversity of multiple taxonomic groups, hereafter multi-taxon biodiversity, are not sufficiently known, and the links between indicators of biodiversity and of management sustainability are not always evident and strong (Oettel and Lapin, 2021). As a matter of fact, multi-taxon biodiversity sampling and analysis is highly demanding in terms of funding, time, and a broad range of expertise and competences (Tomppo et al., 2010). For these reasons, it remains uncertain to which extent, and for which taxonomic groups could forest management for wood supply deteriorate biodiversity compared to unharvested forests, and how SFM can mitigate these effects. The challenges of multi-taxonomic field sampling are being increasingly addressed at the local or regional scales. In Europe, these efforts often consist of exhaustive species censuses across single- or multiple sites to assess the effects of forest structure and management on the diversity of multiple taxonomic groups (see references in Table 1). Table 1 Definitions of the silvicultural systems used in the platform. Silvicultural systems Treatment description Unmanaged No silvicultural interventions applied in the recent past, stands currently under natural development Selection cutting Single-tree and group selection cutting are uniformly distributed across the stand. Shelterwood Overstorey trees in a forest stand are completely removed through multiple progressive cuts designed to promote regeneration making use of the shelter and seed source of remaining trees Clearcutting with retention The forest stand is clear-felled in a single harvesting operation except for solitary trees or tree groups that are deliberately spared Clearcutting The forest stand is entirely harvested in a single operation, resulting in a treeless open area mostly artificially regenerated Coppice with standards The two vertical strata of the forest stand (even-aged coppice as the understorey, and an overstorey of standards which are trees of seedrather than sprouting origin) are harvested respectively by a clearcutting and a selection cutting. Standards can be uneven-aged and the two components have quite different rotation periods. This category also includes the combination of coppice and high forest (i.e. compound coppices) S. Burrascano et al. Biological Conservation 284 (2023) 110176 4 Relatively few examples of such studies can be found in other continents within the temperate and boreal zones (Murray et al., 2017; Bowd et al., 2021; Stokely et al., 2022). The efforts of European researchers to test the effectiveness of SFM reflect the long-lasting and widespread landsharing approach to forest zoning that characterizes Europe, at least since M¨ oller (1922), as compared to regions where a sparing approach is more common (e.g., Australia, North America). However, compared to continent-wide health forest monitoring networks (e.g., International Co-operative Programme on Assessment and Monitoring of Air Pollution Effects on Forests - ICP Forests), research activities related to European forest biodiversity remain uncoordinated at the continental scale and suffer from a lack of harmonization and integration across local studies. Here we aim to i) review the existing information on forest multitaxon biodiversity associated with stand structure and management and ii) identify knowledge gaps in multi-taxon forest biodiversity studies. Starting from this basis, we aim to iii) discuss future research challenges associated with multi-taxon biodiversity in European forests. Ultimately, we intend to encourage new institutional forms of broadscale forest biodiversity data collection and usage to inform forest conservation and management policies in Europe and elsewhere. 2. Methods 2.1. Data collection We aimed at gathering forest biodiversity and stand structural data from many independent research projects and studies on multi-taxon forest biodiversity performed across Europe in the last 20 years. We created a network connecting research groups that collected forest multi-taxon biodiversity data in Europe from local to national scales. The initial network was progressively enlarged by contacting researchers involved in past and ongoing forest multi-taxon biodiversity projects identified through project databases (i.e., LIFE projects’ database: https://webgate.ec.europa.eu/life/publicWebsite/search), publications and personal information. We defined a forest as an ecosystem in which tree cover was equal or greater than 40 % (Sasaki and Putz, 2009). We intended multi-taxon to include simultaneous information on a minimum of three taxonomic groups, representing at least Plantae or Fungi, and Animalia. Stand structure data was defined as tree species composition, volume and size distribution of standing trees (both living or dead), and of lying deadwood when available. To test the comprehensiveness of the platform, we surveyed the research network on relevant literature and gathered 117 published articles (Trentanovi et al., n.d.). We added further articles found through a literature search on ISI Web of Science (accessed on November, 17, 2022 and limited to articles published before 2022) with the formula (TS =(forest AND multi-tax* AND biodiversity)). This search resulted in 130 additional articles, of which 61 describe studies performed in Europe. Among these, 17 focused on non-forest habitats (e.g., wood pastures, urban or agricultural areas), 17 did not fit the multi-taxon or stand structure requirements, five had a different focus (sampling methods, biotic disturbance). Only 22 articles fitted the platform requirements, among which 11 were already listed in the literature gathered through the research network, and five were relative to datasets already included in the platform. Based on this literature assessment we deemed the platform as significantly representative. For each dataset in the platform, we firstly gathered and harmonized the information on sampling designs and protocols (see the ‘protocols’ of Fig. 1) into three ancillary tables that include standardized sampling protocol descriptions of: standing trees, lying deadwood, and biodiversity data. (i.e., ‘protocol data’ in the output section of Fig. 1). The spatial hierarchy of the platform encompasses plots (i.e., delimited forest areas of known geographical coordinates) that are nested into stands (i.e., management spatial units), and stands that are nested into sites (i.e., environmentally homogeneous geographical Fig. 1. Workflow of the platform building process. Blue boxes identify in-progress products; green boxes identify phases of common decisions, brainstorming and comparison with scientific literature; gray boxes indicate data processing; yellow boxes are the outcomes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) S. Burrascano et al. Biological Conservation 284 (2023) 110176 5 areas). We generally refer to sampling units to include both plots and stands, since in a minority of cases (about 15 % of the stands do not have plot IDs but stand IDs), data were sampled at the stand scale, without specific sampling plots. It is also relevant to point out that for most of the collected datasets, several ecosystem components, i.e., taxonomic groups and stand structure, were sampled in the same plots, but in a minority of cases different components were sampled using various designs across a forest stand, allowing for cross-taxon analysis only at the stand scale (see Burrascano et al., 2021 for a thorough discussion of the pros and cons of the two approaches). The studies included in the platform were mainly observational but some experimental studies were also included. In both cases, we collected information on stand age and/or development stage, a categorical definition of the silvicultural system adopted, and many other associated quantitative management data (e.g., time since last harvesting). 2.2. Data management Each original dataset was associated with one data custodian, responsible for data preparation and handling within the platform, and for communication with the dataset contributors. The heterogeneity in sampling designs, measurement methods, spatial scales, target variables and taxonomic groups required the definition of standardized procedures to harmonize raw information and produce a common data structure. We built a relational structure encompassing several tables whose organization and templates derived from an iterative process of proposals and refinements carried out through a wide and open discussion (see connections between boxes in the ‘Input’ section in Fig. 1). The core of the platform (i.e., ‘Sampling unit metadata’ and ‘Raw data’ in Fig. 1) consists of four tables: one containing sampling unit metadata and the others containing the raw data separately for standing trees, lying deadwood, and multi-taxon species composition. The sampling unit metadata includes location, ownership, structure, regeneration and management type. Two key variables within this table are the forest compositional category and the silvicultural system. The former refers to the classification into 14 categories of ecologically distinct forest communities in Europe dominated by specific assemblages of tree species (SI-1). These categories were designed to facilitate the interpretation and communication of indicators on the status and trends of forests in Europe (EEA, 2006; updated by Barbati et al., 2014). The classification into broad silvicultural systems was based on the type of regeneration cut according to Matthews (1989) and refined through an extensive discussion within the network (Table 1). Data were processed in the R programming environment (R Core Team, 2021, R version 4.1.1) using version control to ensure the widest possible hands-on collaboration and data cross-checking. Data inconsistencies that may have originated from data entry errors (e.g., typographical errors), data type storage, species nomenclature, and adherence of datasets to the table structure (e.g., column names, list of possible and plausible values) were qualitatively checked through several validation rules. These (semi)automatic rules were based on data range, length, column reference name, list values, null values, blank values, and data types. After this validation process, data have been corrected or integrated mostly through back-checking to data providers. Species names and higher taxonomic information were extracted from databases and corroborated by experts. All species names were firstly checked using the gnr_resolve() function in the ‘taxize’ package (Chamberlain and Sz¨ ocs, 2013). The species names obtaining scores greater than 0.90 were accepted, while those with lower scores were sent to experts for corroboration. For vascular plants, a further screening was performed through the ‘WorldFlora’ package (Kindt, 2020). Finally, species names that could not be corroborated by experts were checked against the GBIF database (https://www.gbif.org/). Complete taxonomic classification was extracted with the taxonomy() function in the ‘myTAI’ package (Drost et al., 2018). Relationships across tables operate at different spatial scales through univocal IDs for sites, stands, and plots. The templates of the tables for contributing data to the platform are available at: https://www.botto ms-up.eu/en/participants/contribute/contribute-data.html to promote further implementation of the data. Data management is coordinated by a governing board elected by the consortium involving all data contributors according to common bylaws that were discussed and accepted by all the consortium participants (downloadable here: https://www.bottoms-up.eu/en/participants/cont ribute/propose-a-project.html). The bylaws are composed of eight regulation articles partly based on previous experiences of shared datasets (e.g., Biurrun et al., 2019). As it is always the case in the beginning of these sharing processes, a give-and-take approach has been chosen (Kattge et al., 2020; Bruelheide et al., 2019); therefore, joining the consortium is possible for researchers that provide a dataset complying with the bylaws requirements. Data usage is open to anyone proposing a research project involving at least one consortium member by following the instructions at: https://www.bottoms-up.eu/en/partici pants/contribute/propose-a-project.html. A Shiny Web-App (https:// www.bottoms-up.eu/en/results/data-explorer.html) was created to smooth the proposal of projects by allowing for data exploration and filtering based on the sampling unit metadata (SI-2). 2.3. Data analysis and visualization The proportion of sampling units for each compositional category was compared with the share of such categories in the European Union forest area as reported in Barbati et al. (2014). Similarly, sampling unit distribution across broad regeneration strategies (high forest and coppice), and unmanaged areas were compared with the share of forest area under these conditions as reported in McGrath et al. (2015). Sampling unit metadata were visualized through alluvial plots using the ‘ggalluvial’ R-package (Brunson, 2020). The distribution of species and species records across higher taxonomic ranks (phyla and classes) was represented through a phylogenetic tree, encompassing 7335 out of the 8724 species of the platform. The tree was obtained through PhyloT (https://phylot.biobyte.de/) in Newick format and imported in R through the ‘ape’ package (Paradis and Schliep, 2019). To each species in the tree, we associated its higher taxonomic ranks derived through the ‘myTAI’ package (Drost et al., 2018) and the number of occurrences across sampling units (SI-3). The combined information was visualized by using the ‘ggtree’ package (Yu et al., 2017). In addition to the mentioned check for plausible ranges of values, structural data were subjected to specific integration processes. Heights of standing trees were integrated by means of height-diameter relationships (hypsometric models); whereas deadwood fragments measurements were integrated through data imputation performed using the ‘mice’ package (van Buuren and Groothuis-Oudshoorn, 2011). Data integration was performed individually for each dataset by applying the predictive mean matching (PMM), i.e., assessing imputation uncertainty through the examination of the variation in imputed values when treated as real, and including forest compositional categories and types, and spatial variables (plot and site). Few datasets lacked any measures of height/length and were integrated by using the whole data platform. A total of 42,643 tree heights out of 178,098 were calculated by means of hypsometric models; and 5011 diameters and 9317 lengths were imputed out of 58,824 lying deadwood fragments. Based on these data, the distributions of sampling unit mean of Diameter at Breast Height (DBH) for standing trees and of diameter for lying deadwood were calculated (SI-4, SI-5). Biodiversity and stand structure indices may be related to environmental conditions by using parameters published in the framework of other research projects. For instance, each sampling unit was spatially associated with data on soil characteristics that were obtained from the S. Burrascano et al. Biological Conservation 284 (2023) 110176 6 European topsoil physical properties map (Ballabio et al., 2016). Among the multiple soil properties available in this dataset, those currently linked to each forest multi-taxon sampling unit over a 1000 m buffer from the center of the sampling unit are: Available Water Capacity, Bulk density (derived from soil texture datasets), Soil textural classes derived from clay, silt, and sand maps. Similarly, climatic data were obtained from CHELSA v.2.1 (Karger et al., 2017), at 1000 m resolution. Bioclimatic variables were derived as long-term means or sums over the 1981–2010 period, and included mean annual temperature, annual range of air temperature, annual precipitation amount, and precipitation seasonality. Each sampling unit is also associated to a heat load index (HLI) expressing the topographic influence on incident solar radiation (McCune and Keon, 2002). 3. Results 3.1. Overview of the existing data 3.1.1. Sampling units A total of 3591 sampling units across 220 sites in 12 European countries were gathered (SI-2, https://zenodo.org/record/7886698#. ZFEs7HZBxD8), ranging from Sweden to southern Italy in latitude, and from France to Lithuania in longitude (Fig. 2). The harmonization involved 34 local datasets (Table 2) and 185 researchers. In general, no clear pattern of association appears between silvicultural systems and forest compositional categories. For instance, shelterwood is applied to almost all compositional categories (Fig. 3); even though coppice with standards were associated mostly with mesophytic and thermophilous deciduous forests. Importantly, most sampling units (94 %) were referred to a habitat type of conservation concern according to the European cornerstone of biodiversity conservation, the Habitats Directive (92/43/EEC). The highest representation (82 %) was found for the forests of temperate Europe (group 91). About a quarter (27 %) of the total number of sampling units referred to as priority habitat types, with, in order of decreasing frequency, 91H0*- Pannonian woods with Quercus pubescens, 91G0*- Pannonic woods with Quercus petraea and Carpinus betulus, 91E0*- Alluvial forests with Alnus glutinosa and Fraxinus excelsior (Alno- Padion, Alnion incanae, Salicion albae). 3.1.2. Taxonomic information The dataset comprises a wide range of taxonomic groups across the kingdoms of fungi, plants, and animals (SI-3, https://zenodo. org/record/7886698#.ZFEs7HZBxD8), with 8724 species, 2979 genera, 729 families, 193 orders, 44 classes, and nine different phyla (Fig. 4). The taxonomic groups originally considered in each study include heterogeneous taxonomic ranks, from kingdom to order, and, in some cases, include only specific morphological or ecological groups commonly used in sampling and identification (e.g., macrofungi, epiphytic lichens, saproxylic beetles). Most plots have information on four or more different taxonomic groups, with an average of 4.6 groups per plot. Some plots have information on only one or two taxonomic groups, but on at least three taxonomic groups at the stand level, as required by the platform bylaws. The classes that are represented by the highest number of species are also those represented by the highest number of records (Fig. 5), i.e., Fig. 2. Distribution of the platform sampling sites in Europe. Gray areas are covered by forests with a tree cover greater than 40 % according to the European Forest Institute Forest Map of Europe (Kempeneers et al., 2011). Number of taxa are represented by color, while number of sampling units by dot size. S. Burrascano et al. Biological Conservation 284 (2023) 110176 7 Insecta (3244 species across 88,338 records), Agaricomycetes (2077 species across 44,418 records), Magnoliopsida (1182 species across 71,458). However, this pattern differs for Bryopsida, which are represented by only 280 species in a very high number of records (27,551). This means that some Bryopsida species occur in a very high number of sampling units; for instance, Hypnum cupressiforme is the species occurring in the highest number of sampling units (5082) among all species in the platform (see also Blasi et al., 2010). On the other hand, Insecta, Magnolipsida and Agaricomycetes are the most represented classes among species singletons (i.e., species occurring only once in the platform) with respectively 35 %, 29 % and 13 % of sinlgetons vs. less than 3 % for Bryopsida. 3.1.3. Structural attributes Being required by the bylaws, diameters of standing trees are available for all sampling units, either at the plot or stand scale, with a total 2889 sampling units, with additional data on tree height, deadwood and tree-related microhabitats in respectively 2356, 2309, and 1388 plots. The mean diameters of standing trees (both living and dead) and lying deadwood within each sampling unit across the platform vary across silvicultural regimes. The former shows a left-skewed distribution for clearcutting and clearcutting with retention and a bimodal distribution for coppice with standards. The lying deadwood mean diameters have a peak at lower values if compared with standing tree DBH distributions, and often show a wider range, in some cases with a bimodal distribution (Fig. 6). 3.2. Gaps in knowledge 3.2.1. Sampling units Sampling units with multi-taxon biodiversity data associated with stand structure and management information (SI-2) are available for all the 14 European forest compositional categories (Barbati et al., 2014) although unevenly distributed among them. European beech forests are over-represented with respect to the area they occupy (Fig. 5A), with lowland and mountainous beech forests representing more than 55 % of the sampling units (Barbati et al., 2014) (Fig. 5A). The distribution of sampling units across management systems, i.e., timber harvesting relying on resprouting (coppice) or seed regeneration (high forest) or no harvesting (unmanaged), differed from their area extent (McGrath et al., 2015) only for coppices (Fig. 5B). Relevant gaps remain for some crucial management information, for instance type and year of last intervention were not available for about 40 % of sampling units. This indicates that detailed management information is often not available to forest biodiversity researchers. However, the sampling unit metadata allow a close look into the management and composition of the forests included in the platform. Most of the available sampling units are within public forests, naturally regenerated, with shelterwood and selection cutting systems being by far the most represented, and single- and multi-storied forests being similarly frequent (Fig. 3). 3.2.2. Taxonomic information Some taxonomic classes are underrepresented both in terms of species and records, and this is especially true for fungi other than Agaricomycetes. For instance, Eumycetozoa, Mucoromycetes and Pucciniomycetes are represented by only one species that occurs in less than 5 sampling units. The rarity of these classes is even more striking if we account for the fact that fungi were among the most commonly sampled groups of organisms. 3.2.3. Structural attributes Notwithstanding differences in the forest definition adopted and in DBH threshold, the diameter distributions of the data within the platform reflect the distribution of growing stock across diameter classes at the European level (Forest Europe, 2020), with most sampling units having a mean DBH between 20 and 40 cm (Fig. 6). On the other hand, in European multi-taxon studies, the share of trees over 40 cm is higher than the share below 20 cm. This means that multi-taxon studies are partly biased towards mature developmental phases. On the one hand, this may result in an underrepresentation of species related to different phases of the forest succession (Hilmers et al., 2018), especially of the earliest; on the other hand, it may contribute to demonstrate the links between large trees and forest biodiversity (Koz´ ak et al., 2023). 4. Discussion 4.1. Solids and voids of the existing data This study is the first attempt, both in Europe and globally, to encompass extensive and comprehensive information on forest management, structural attributes, and multi-taxon biodiversity in a single, harmonized and publicly explorable platform. Although we collected most data deriving from forest multi-taxon studies performed in Europe, the resulting data is unevenly distributed across compositional categories. Further efforts should thus focus on attaining a good representation for boreal, hemiboreal, thermophilous and Mediterranean compositional categories, as well as for mire, swamp, and floodplain forests. Among these, boreal and hemiboreal forests are widespread in Europe, thus their underrepresentation in the platform is a clear knowledge gap. Other categories, instead, have a limited extent across Europe, with floodplain forests displaying the lowest cover (Barbati et al., 2014). Although floodplain forests are Table 2 General descriptors of the local datasets. Dataset ID: ID of dataset, Country: country where the study has been made, N sites: number of sampled sites, N units: number of sampling units, Reference: main literature references of the 34 datasets collected for this work. Dataset ID Country N sites N units Reference BE_PS1 BE 2 32 De Smedt et al., 2019 BE_PS2 BE 1 53 de Groote et al., 2017 BE_KV1 BE 5 462 Vandekerkhove et al., 2016 CH_TL CH 1 69 Haeler et al., 2021 CZ_JH1 CZ 6 106 Hofmeister et al., 2019 CZ_JH2 CZ 1 230 Hofmeister et al., 2013 CZ_MR CZ 1 45 Chamagne et al., 2016 DE_ID DE 1 526 Doerfler et al., 2017 DE_JP DE 1 135 Storch et al., 2020 DE_PS DE 3 150 Schall et al., 2018 DK_JC1 DK 6 40 Lelli et al., 2019 DK_JC2 DK 2 107 Mazziotta et al., 2016 DK_JC3 DK 1 30 – DK_SK DK 16 386 Byriel et al., 2020 FR_AM FR 12 33 Cocquelet et al., 2019 FR_JP FR 3 70 Janssen et al., 2018 FR_NK FR 35 43 Korboulewsky et al., 2021 FR_YP FR 24 300 Paillet et al., 2015 GR_FX GR 1 4 – HU_FT HU 1 36 Horv´ ath et al., 2023 HU_PO1 HU 1 35 Tinya et al., 2021 HU_PO2 HU 1 30 Elek et al., 2018 HU_RA HU 8 22 – IT_AC IT 3 18 Cutini et al., 2021 IT_EA IT 6 199 D’Andrea et al., 2016 IT_SB1 IT 1 36 Blasi et al., 2010 IT_SB2 IT 2 66 Sabatini et al., 2016 IT_TS IT 2 20 Sitzia et al., 2017 LT_GB LT 20 143 – SK_DK SK 3 18 Kameniar et al., 2021 SK_MM SK 3 22 Koz´ ak et al., 2021 SK_MS SK 3 18 Langbehn et al., 2021 SK_MU SK 1 65 Ujh´ azy et al., 2018 SW_BN SW 25 50 G¨ otmark, 2013 S. Burrascano et al. Biological Conservation 284 (2023) 110176 8 considered as biodiversity hotspots (Przepi´ ora and Ciach, 2022), these habitats are nowadays not only rare in European landscapes, but occur in settings that have been profoundly altered by humans, thus characterized by high habitat fragmentation and low ecosystem integrity. For these reasons, floodplain forests should be primarily addressed by biodiversity studies, especially in view of restoration actions (Dufour et al., 2019). On the other hand, the share of sampling units referred to as priority habitat types (27 %) is higher than what is reported in terms of area, i.e., 23 % according to European Commission (2019), showing a great potential of the existing data in the assessment of the conservation status and relation with management of forests of primary conservation concern in Europe. The distribution of the existing data across silvicultural systems shows a tendency among researchers to perform multi-taxon biodiversity studies in forests that are perceived as less intensively managed, such as those under selection or shelterwood management regimes. Clearcutting is represented mainly in plots related to experimental studies testing novel harvesting practices as this silvicultural system is deemed as strongly threatening forest depending species (Savilaakso et al., 2021). Unmanaged forests have often been sampled as a relevant reference in comparison to managed forests, especially if strategies that are generally perceived as sustainable are applied (Paillet et al., 2010). However, it should be noted that the unmanaged sampling units in the platform may not be associated with old-growth condition since they vary widely in terms of time passed since the last management intervention, i.e., from 20 to more than 100 years, and these differences have to be accounted for when contrasting managed and unmanaged forests within the platform. The distribution across unmanaged, coppiced and high forests confirms that coppicing systems are relatively understudied, particularly for multi-taxonomic biodiversity. This may be partly related to the perception of these forests as less relevant for biodiversity and related ecosystem services. Nevertheless, some studies suggest the opposite (H´ edl et al., 2010) and demonstrated a certain degree of association of species of conservation concern with actively coppiced stands (Kosulic et al., 2016). This uneven distribution may also be related to the progressive reduction of coppiced forests in Europe, which are gradually being actively or passively converted into high forests, or simply abandoned (Burrascano et al., 2017). In general, the lack of multi-taxon biodiversity information from coppices represents a knowledge gap for supporting policy decisions on coppice forest management, which is especially relevant in view of their renewed prominence in climate adaptation policies and forest multifunctionality (Cutini et al., 2021). 4.2. Research potential and working hypotheses Conversely to other data sharing platforms focusing on individual aspects of ecosystems, e.g., vegetation (Bruelheide et al., 2019), forest multi-taxon studies put in place different expertise and data on the three main components of ecosystems: species composition, structure and function. As such, these data collectively have the potential to promote interdisciplinary studies and to unveil the outcomes of different conservation and management policies on the biodiversity of multiple taxonomic groups as mediated by structural stand features, e.g., the diameter distributions of standing trees and lying deadwood. 4.2.1. Harmonizing methods and schemes The joint assessment of existing forest multi-taxon biodiversity data already stimulated a harmonization effort for sampling protocols (Burrascano et al., 2021). Similarly, a harmonized platform may serve as a Fig. 3. One-, two-, and multi-storied forest sampling unit distribution across type of ownership, regeneration, silvicultural system (cws: coppice with standards), forest type, and habitat type. Vertical blocks represent clusters of sampling units for which the same condition (e.g., natural regeneration) occurs, with block height depending on the number of sampling units for which that condition occurred across structural types (single-, two- and multi-storied). Flows between the blocks show the combination of values for different structural types (e.g., number of one-storied plots within a public property originated from planting). In the forest type column, left numbers refer to the forest categories as reported in SI 1 and in EEA (2006), while the number after the dot refers to a specific type within that category (please refer to EEA, 2006). Forest types represented by less than 50 sampling units are identified by stars (* 3.1; ** 4.2; *** 7.4, 7.5; **** 8.1, 8.2, 8.7, 8.8, 9.1, 10.1, 11.1; ***** 11.3, 11.4, 12.1, 13.2, 14). In the Habitat column, sampling units are gathered in groups of habitat types divided among those having priority (followed by “*”) or not, and referred to the codes: 90: Forests of Boreal Europe, 91: Forests of Temperate Europe, 92: Mediterranean deciduous forests, 93: Mediterranean sclerophyllous forests, 94: temperate mountainous coniferous forests, 95: Mediterranean and Macaronesian mountainous coniferous forests. S. Burrascano et al. Biological Conservation 284 (2023) 110176 9 Fig. 4. Phylogenetic tree of the species enclosed in the data platform (7335 out of the 8724 could be included). The colored sectors refer to phyla; external bars show the log-transformed number of records for each species; gray circular sectors show the representation of the classes for which more than 50 species occur. Fig. 5. A: Share of the number of sampling units and of forest area across different forest compositional categories (A) in EU-28 based on Barbati et al. (2014); and across two broad methods of regeneration (coppice and high forest) and the absence of silvicultural intervention (B) based on McGrath et al. (2015) . S. Burrascano et al.