The BenBioDen database, a global database for meio-, macro- and megabenthic biomass and densities

Benthic fauna refers to all fauna that live in or on the seafloor, which researchers typically divide into size classes meiobenthos (32/64 µm–0.5/1 mm), macrobenthos (250 µm–1 cm), and megabenthos (>1 cm). Benthic fauna play important roles in bioturbation activity, mineralization of organic matter, and in marine food webs. Evaluating their role in these ecosystem functions requires knowledge of their global distribution and biomass. We therefore established the BenBioDen database, the largest open-access database for marine benthic biomass and density data compiled so far. In total, it includes 11,792 georeferenced benthic biomass and 51,559 benthic density records from 384 and 600 studies, respectively. We selected all references following the procedure for systematic reviews and meta-analyses, and report biomass records as grams of wet mass, dry mass, or ash-free dry mass, or carbon per m2 and as abundance records as individuals per m2. This database provides a point of reference for future studies on the distribution and biomass of benthic fauna.

(2020) 7:206 | https://doi.org/10.1038/s41597-020-0551-2 www.nature.com/scientificdata www.nature.com/scientificdata/ slope failure 8 . It can be ecosystem engineers, i.e., organisms that alter the physical environment to change directly or indirectly the availability of resources to other organisms 9 , and modify hydrodynamics. Megabenthos or fauna larger than 1 cm includes organisms such as scleractinian corals or sponges that form biological structures and Full-text records excluded for presenting experimental studies (n = 23), for inaccessibility of publications or data (n = 10), for averaging data over several sampling stations instead of presenting the data per individual station (n = 53), for presenting density not as ind. m -2 or in a unit that could be converted to ind. m -2 (n = 39), for presenting densities of specific taxa and not of the whole size class (n = 31), for reporting meta-studies or reviews and not primary research (n = 14), for presenting models (n = 8), for lacking details about geographic location (n = 17), for presenting fauna attached to a whale fall (n = 1), for combining densities of invertebrate and vertebrate megafauna (n = 2), for combining several size classes (n=2), and for not reporting meiobenthic, macrobenthic, and/ or invertebrate megabenthic densities (n = 153). Full text records excluded for presenting experimental studies (n = 7), for inaccessibility of publications or data (n = 4), for averaging data over several sampling stations instead of presenting the data per individual station (n = 13), for presenting benthic biomasses not as wet weight, dry weight, ash-free dry weight, or carbon per m 2 (n = 7), for presenting biomasses of specific taxa and not the whole size class (n = 17), for reporting meta-studies or reviews and not primary research (n = 7), for presenting models (n = 3), for lacking details about geographic location (n = 3), for presenting benthic biomass as ranges instead of mean or median values (n = 1), for combining vertebrate and invertebrate megafauna biomass data (n = 1), and for not reporting benthic biomasses (n = 47). Full-text studies assessed for eligibility (n = 494) a b Fig. 1 Flow chart explaining how the database was compiled. It shows how publications and datasets were identified and which selection criteria were used to exclude studies from the final "BenBio" part of the BenBioDen database (panel a) and the final "BenDen" part of the BenBioDen database (panel b).

Studies included in the
thus provide new habitats for associated fauna 10,11 . Other examples of megabenthos assemblages are oyster reefs and mussel beds that create biogeochemical hotspots for the burial of organic matter and the recycling of nutrients [12][13][14] . Additionally, mussels, cockles, oysters, but also sea cucumbers, are part of the human diet. Despite their ecological importance, benthic ecosystems face increasing pressures from fishing, pollution and litter disposal, gas and oil exploration, extraction of minerals, development of coastlines, shipping, tourism, invasive species, and wind farms [15][16][17] . Sea-level rise can force intertidal habitats, such as salt marshes and tidal flats, to migrate landwards where they may be squeezed against artificial coastal structures 18 . This "coastal squeeze" leads to the loss of intertidal habitats and macrobenthic biomass 19,20 . Furthermore, ocean acidification will strongly impact tropical and cold-water coral reefs [21][22][23] and calcareous fauna such as bivalves, gastropods, bryozoans, echinoderms, and foraminifera 24 . A combination of changes in pH, temperature, and oxygenation will even affect the export flux of particulate organic carbon (POC) to the seafloor 25 and subsequently result in decreased benthic biomass 26 .
Evaluation of the severity of these threats and climate change for the benthic ecosystem on a global scale requires quantifying the role of benthos and its biomass and density in particular. Here, we introduce the open access "BenBioDen database" 27 that, in comparison to previous databases by, e.g., Rex et al. 28 and Wei et al. 29 , makes the benthic biomass and abundance records freely available and describes the data selection procedure transparently. Furthermore, this database includes records from the whole globe and not only from specific  Table 1. References of biomass conversion factors to calculate metazoan meiobenthic, macrobenthic, and megabenthic biomasses as wet mass (WM), dry mass (DM), ash-free dry mass (AFDM), and C content (C). The "BenBioDen database" reports 1,445 benthic biomass and 2,085 benthic density studies and datasets identified following standardized procedures for systematic reviews and meta-analyses 30 . As a result, we extracted 11,792 georeferenced records of benthic biomass (1,240 metazoan meiobenthos records, 9,292 macrobenthos www.nature.com/scientificdata www.nature.com/scientificdata/ records, and 1,260 invertebrate megabenthos records) and 51,559 georeferenced records of benthic densities (4,129 metazoan meiobenthos records, 46,389 macrobenthos records, and 1,041 invertebrate megabenthos records) from 384 and 600 selected studies, respectively. We report benthic biomass as g wet mass (WM) m −2 , as g dry mass (DM) m −2 , as g ash-free dry mass (AFDM) m −2 or as g carbon (C) m −2 . All biomass and density data records include further information about the mesh size used to separate meiobenthos from macrobenthos and megabenthos, and macrobenthos from megabenthos, and the sampling gear. In this way, researchers can decide whether they wish to exclude specific studies that do not match organism size criteria or sampling gear criteria. The database provides an important point of reference for future studies on the distribution and biomass of benthos and may also stimulate future sampling campaigns by indicating undersampled locations and water depth.

Methods
In April and May 2019, we compiled the "BenBio" part of the "BenBioDen database" following the "Preferred Reporting Items for Systematic reviews and Meta-Analyses" (PRISMA) Statement for systematic reviews and meta-analyses 30 (Fig. 1a). In the first PRISMA step, the "Identification" step, we identified 1,373 articles in the Web of Science using the key words "marine meiofauna biomass", "marine macrofauna biomass", "marine megafauna biomass", "marine meiobenth* biomass", "marine macrobenth* biomass", "marine megabenth* biomass", "nematode biomass", and "benthic 'standing stock'". We located an additional 201 publications based on expert knowledge. A search of the PANGAEA (R) Data Publisher (https://www.pangaea.de/) identified 1,488 datasets representing 148 publications using the key words "meiofauna biomass", "macrofauna biomass" and "megafauna biomass". Further 30 datasets were found in the EOL data archive (http://data.eol.ucar.edu/), through citations in review papers, and based on expert knowledge. After removing duplicates, we screened the titles and abstracts of 1,445 studies (Online-only Table 1) in PRISMA step 2 ("Screening"; Fig. 1a). This step excluded 951 studies because they did not report biomass values. In the Eligibility step (step 3; Fig. 1a), we assessed full texts of 494 studies for eligibility and excluded 110 studies because they did not report biomass, the publications or data were not accessible, or they did not report benthic biomass in appropriate units (g WW m −2 , g DW m −2 , g AFDW m −2 , g or mol C m −2 ). Further reasons for excluding full texts included combining benthic biomass for several size classes, reporting benthic biomass for particular taxa rather than the whole size class, presenting biomass for faunal assemblages and/or a group of sampling stations rather than for individual stations, not presenting primary research or lacking geographical details about sampling stations. We also excluded studies that estimated benthic biomass using modelling approaches, that conducted manipulative experiments, or did not report benthic biomass as single values, means or median values, but instead as ranges. The final "BenBio" part included 384 studies from which we extracted 11,792 georeferenced benthic biomass entries (Online-only Table 1; Fig. 1a).
For 12% (BioBen part) and 4% (BioDen part) of all data records, no exact sampling location in geographical coordinates (latitude, longitude) was indicated. For these cases, we approximated the coordinates of the sampling locations using Google Maps based on information about sampling area or based on maps presented in the original publications. We labelled these data records as 'approximated location' .
For studies that presented biomasses in several units, such as WM and DM, we report the data only once (preferred units: WM > DM > AFDM > C). The authors of this study intended to report all data records in the 'raw' units in which benthic fauna was measured initially. Whenever unknown conversion factors precluded calculating biomass back to 'raw' units, we noted this issue in the database using the label 'converted data' and listed references for the individual biomass conversion factors in the database. Furthermore, we prepared Table 1 that reports all literature used by the authors of the original studies to convert their biomasses size-class dependent to WM, DM, AFDM, and C content.
The authors of the various studies compiled in this database sometimes used different lower and upper limits (in mm) for mesh sizes of nets and/or sieves to define the size class. Whenever an original study reported a lower and/or upper limit mesh size, we included this information in the database as 'sieve mesh size (mm) lower limit' and 'sieve mesh size (mm) upper limit' . Studies lacking this information were scored as NA.
For those studies that reported data as mean or median ± error terms, we incorporated only mean or median values into the database. In all cases that did not report benthic biomasses and/or densities in the text or in tables, but presented them in figures, we extracted biomass and/or density values from these figures using ImageJ 31 .

Data Records
The BenBioDen database is openly accessible in the Dryad Digital Repository 27 and includes two txt.files, i.e. the List of studies for BenBio database file and the List of studies for BenDen database file, and two csv.files, i.e., the BenBio database file and the BioDen database file. The List of studies files list all 3,531 studies alphabetically (benthic biomasses: 1,445 studies; benthic densities: 2,086 studies) which we identified in the "Identification" www.nature.com/scientificdata www.nature.com/scientificdata/ step of the systematic review after removing all duplicates. Each data entry in the BioBenDen database contains information about the region where the biomasses and/or densities were sampled and the corresponding ocean, the geographical location (latitude, longitude), whether geographic location was exactly known or approximated, water depth (in m), and a depth range following Dunne et al. 32 . Dunne and co-authors divided the ocean in near-shore areas that stretch to 50 m water depth, continental shelves from > 50 to 200 m water depth, continental slopes from > 200 to 2,000 m water depth, and continental rises/abyssal plains > 2,000 m water depth. The database indicates whether we determined the biomasses as WM, DM, AFDM, or C content; densities are reported as ind. m −2 . The database also reports the specific size class (metazoan meiobenthos, macrobenthos, invertebrate megabenthos), the mesh size of the sieves used by the authors of the studies to separate the different size classes and the sampling gear.
Meiobenthos biomasses were quantified mostly on the continental slope (35%) and on the continental rise and abyssal plains (31%) that collectively encompass 95% of the ocean seafloor 32 (Fig. 5). In contrast, near-shore areas (29%) and continental shelves (21%; Fig. 5) dominated macrobenthic biomass samples, although these areas collectively encompass < 5% of the global seafloor 32 . In 35% of the cases no sampling depth was given in the original publications. Also 47% of all megabenthos biomass records came from areas < 50 m water depth, whereas only 10% of all megabenthos biomass samples were taken in the largest part of the seafloor, the continental rise and abyssal plains (Fig. 5). Hence, not surprisingly the benthic biomass database is biased towards shallow waters (<200 m) in the northern hemisphere, particularly, in the North Atlantic.
Meiobenthic density samples were mainly taken in the Atlantic Ocean (including the Gulf of Mexico and the Mediterranean Sea; 59%) and in the Pacific Ocean (22%) (Figs. 6, 7), whereas macrobenthic density was dominantly sampled in the Atlantic Ocean (including the Gulf of Mexico and the Mediterranean Sea; 87%) (Figs. 6, 7). Megabenthic densities originated from the Arctic Ocean (53%), the Atlantic Ocean (including the Gulf of Mexico and the Mediterranean Sea; 26%), and the Pacific Ocean (14%) (Figs. 6, 7). More than 83% of all samples were taken in the northern hemisphere (>1°N), in case of macrobenthos, even 98% of all density samples were taken > 1°N (Fig. 8).
Meiobenthic and megabenthic densities were sampled to 65% and 56% at the continental slope and at the continental rise and abyssal plain (Fig. 7), whereas information about sampling depth was missing for 82% of the macrobenthos samples that originated predominantly from the North Atlantic. When these records are not taken into account, most of the macrobenthic density samples were collected in near-shore areas (38%) and at the continental shelf (33%). Hence, benthic density samples are biased towards the northern hemisphere and in particular towards the North Atlantic and the Arctic Ocean (Fig. 4).
Differences in size ranges of meiobenthos, macrobenthos, and megabenthos. Metazoan meiobenthos usually includes organisms that pass through 500 μm to 1 mm mesh size and are retained on sieves with 44 μm mesh size 1 , though deep-sea biologists often use a lower mesh size limit of 32 μm for metazoan meiobenthos 34 . In our database, however, the lower mesh size limit for metazoan meiobenthos ranges from 20 μm to 74 μm, and the upper mesh size limit spans from 100 μm to 2 mm because of the different mesh sizes chosen by the authors of the original studies. Hence, some metazoan meiobenthos records include organisms that might be allocated to microbenthos, and other records that group them with macrobenthos.
Macrobenthos refers to organisms retained on a mesh of 0.5 cm, though different studies used mesh sizes between 0.5 mm and 2 mm 35 . In our database, however, authors of different studies sieved macrobenthos samples with meshes ranging from 0.25 mm to 20 mm in size. This implies, that depending on the size range used for macrobenthos, some macrobenthic records might include also be metazoan meiobenthos.
Invertebrate megabenthos are larger than macrobenthos and defined as invertebrates visible in bottom photographs (> 1 cm or > 3 cm 36 ). Most megabenthic biomass and density records in the BioBenDen database lack specific information about minimum size (82% of all megabenthic biomass records and 79% of all megabenthic density records), but the studies that report a minimum size used a minimum animal length between 0.5 cm and 2 cm. Consequently, part of the megabenthic biomass and density data unavoidably might include some macrobenthos.
Therefore, researchers should consider the lower and upper sieve mesh sizes when using data from this database to ensure that the data coincide with their size requirements.