Mesopelagic Mesozooplankton and Micronekton Database

Abstract

The Mesopelagic Mesozooplankton and Micronekton Database (MMMD) compiles quantitative data on the distribution and density of mesopelagic (200–1000 m) mesozooplankton and micronekton (0.2–20 mm and 20–200 mm) species, using 258 published and unpublished sources spanning from 1880 to 2016. This extensive dataset includes 266,611entries, covering a broad temporal, spatial range and diel changes with varying levels of completeness. The data were standardized to address inconsistencies in sampling methods, mesh sizes, and taxonomic classifications. Despite some limitations, including gaps in spatial and depth coverage, the database provides a valuable resource for the mesopelagic macro-ecological research, which is updatable. Comparisons with existing databases underscore its unique contributions to the study of mesopelagic ecosystems.

Background & Summary

The changing climate and rapid population growth with increasing demand for protein have led to a decline in shelf and slope fish stocks. ‘New resources’ to unlock the economic potential of the global oceans can be found in the mesopelagic realm¹ in the form of the large unexploited biomass of midwater fish, krill and cephalopods². However, knowledge of the mesopelagic fauna abundance, biology and ecosystem role in the open-ocean system is inadequate to reliably assess the potential impacts of mesopelagic resources harvesting. The lack of such knowledge hinders the ability to assess the oceanic response and feedback in times of drastic climate change and increased human activities³. Thus, understanding the global distribution of mesopelagic mesozooplankton and micronekton is critical for assessing their ecological roles and potential as future marine resources. However, current knowledge is limited due to inconsistent data coverage, particularly in the mesopelagic zone, and a lack of standardized methodologies across available datasets.

For instance, while the Ocean Biogeographic Information System (OBIS) contains over 31.3 million records of more than 120,000 marine species: >50% come from the epipelagic zone and <20% are from the mesopelagic realm^4,5. Due to the wide depth range of the mesopelagic zone, records are spatially and vertically inconsistent, making quantitative analyses challenging⁶. Webb et al. found OBIS records decrease tenfold below 200 m, with another order of magnitude drop at 1,000 m (Figure S1), confirming a lack of consistent global data⁵. Plotting the horizontal distribution of OBIS records against depth (Figure S2) reveals a sampling bias, with almost no samples from ocean basin centers, and this gap widens with depth. Sampling is also skewed toward the North Atlantic, with minimal efforts in the Indian, Arctic, and Southeast Pacific Oceans, and these data gaps increase with depth. OBIS data often lack AphiaIDs, unique identifiers for marine species in the World Register of Marine Species⁷ (WoRMS), which are essential for accurately filtering and cross-referencing species across different datasets. Without AphiaIDs, data filtering by species or phylum becomes inconsistent, making it harder to ensure accurate species identification. Additionally, OBIS records focus on species occurrences rather than biomass or abundance, and uneven sampling efforts can skew distribution patterns. Constantly changing taxonomies complicate standardization, and the data often emphasize adult organisms, leaving gaps in zooplankton life stage coverage.

Several databases provide information on marine species, but each has limitations, particularly for the mesopelagic zone. The COastal & Shelf Plankton and Oceanic Biogeographic Observation Database⁸ (COPEPOD) offers quantitative data on different species and communities, but it contains fewer records for the mesopelagic zone and remains incomplete. The Marine Ecosystem Biomass Data⁹ (MAREDAT) is the first inventory based on individual organism counts, focusing on meso- and macrozooplankton. However, its data are scattered across multiple formats, making extraction difficult, and there is no option to filter data by depth strata. The A database of biodiversity time series for the Anthropocene¹⁰ (BioTIME), which includes millions of records, has limited relevance to mesopelagic species, with only ‘Fish’ and ‘Marine invertebrates’ categories available for mesopelagic mesozooplankton and micronekton. Although these datasets offer biomass and abundance estimates, the lack of depth specificity complicates data interpretation, and most estimates pertain to the epipelagic zone.

To the best of our knowledge, no work has attempted to synthesize the available global estimates of the mesozooplankton and micronekton community to explore the current state of knowledge or to perform global comparisons between geographical areas. Such syntheses were only attempted on total epipelagic (0–200 m) mesozooplankton biomass or are generally outdated^11,12. Micronekton, especially mesopelagic fishes, is a better studied group of organisms due to its high biomass and increased interest in potential harvesting as a replacement of depleted coastal fish stocks. Most of the work on mesopelagic mesozooplankton is applicable to either ocean basins or even smaller features (seamounts, eddies, etc.)^13–16 and is generally taxon-specific^17–20. Available global overviews of mesopelagic mesozooplankton and micronekton are listed in Table 1.

Table 1.

Summary of available reviews/syntheses of mesopelagic organisms, aspects being studied, and regions covered.

Organism studied	Aspect being investigated	Region	Ref.
Plankton plants and animals	Vertical distribution	Global	30
Holopelagic animals, mesopelagic larvae	Vertical distribution in relation to environmental factors	Global	31
Mesopelagic fishes	Distribution, ecology, and life history	Global	2
Mesopelagic fish (Myctophidae)	Trophic relationships	Southern Ocean	32
Mesozooplankton	Community respiration	Global	33
Mesopelagic micronekton (mesopelagic fishes, micronekton and cephalopods)	Biogeography	North Pacific	34
Zooplankton communities	The baseline estimates of major structural parameters and seasonal dynamics, abundance, and biomass	Arctic Ocean	35
Mesozooplankton and micronekton	Distribution of biomass	9 Provinces	21
Zooplankton	Overview of the diversity and basic biology with an emphasis on abundance, distribution and feeding	Southern Ocean	36
Pteropods	Global distribution of carbon biomass, with a particular emphasis on temporal and spatial patterns	Global	37
Mesozooplankton	Distribution of biomass	Global	38
Deep-sea cephalopods	State of biogeographical knowledge	Global	39
Mesozooplankton	Spatial and temporal patterns from Discovery Investigations	Southern Ocean	40
mesopelagic shrimps (Sergestidae)	Global distribution	Global	41
Siphonophorae (Cnidaria: Hydrozoa)	History of discovery, species richness, a summary of worldwide distribution and some biological aspects	Global	42

The main goal of this work is to summarize the current state of knowledge on the global distribution of mesopelagic mesozooplankton and the micronekton community through the following steps:

1. compiled all available data from the literature and unpublished datasets on mesopelagic species distribution and abundance to create the first mesopelagic database;

2. systematizing and standardizing available information to account for differences in sampling depth, gear type used, and seasonality of sampling;

3. building a unique, citable reference database for the compiled datasets.

Methods

This work builds upon the research of McIvor²¹, who during his MSc project compiled zooplankton distribution data from various sources to create vertical distribution profiles for organisms in the epi- and mesopelagic layers of nine Longhurst’s²² pelagic ecological provinces. McIvor’s database included 71 articles and 220 study locations and allowed for inter-area comparisons during the summer period. This work expands on previous efforts by attempting to gather as comprehensively as possible available literature on abundance and biomass of mesopelagic zooplankton and micronekton during various seasons.

To expand the intial database, a literature search was conducted via the keywords: ‘mesopelagic,’ ‘vertical distribution,’ ‘mesopelagic zooplankton,’ ‘mesopelagic micronekton,’ ‘twilight zone,’ and ‘deepwater plankton’ through the Web of Science and Google Scholar search engines. Any paper containing data on mesopelagic mesozooplankton biomass or abundance were selected. Additional sources included unpublished data from Dr. A. Yamaguchi (Hokkaido University, Japan) on copepod data from the North Pacific²³ and Dr. Evgeny Pakhomov’s (University of British Columbia, Canada) unpublished data from the Southern Ocean (Figure S3a). From each article, raw density data for zooplankton taxa were collated on the basis of biomass and/or density at discrete depth ranges within the mesopelagic layers (200–1,000 m). In addition to density data, vertical distributions and geographical locations, information related to the date of collection, sampling location, net mesh size (or sieve mesh size for bottle data) and taxa was also recorded. Samples from epipelagic and/or bathypelagic zones were also included in the database if the sampling is performed beyond the mesopelagic zone. Out of many entries collected for the mesopelagic, only data that had quantitative measurements of biomass and abundance were included in the database. Original columns of depth ranges and abundance/biomass estimates are preserved in the database.

In addition, we extracted relevant biomass and abundance data from the PANGAEA Data Publisher²⁴ for records with available geolocation and depths greater than 200 m, specifically targeting zooplankton and micronekton. This process resulted in 38,017 entries sourced from 25 studies.

We also downloaded abundance, biomass, and composition data from the COPEPOD global plankton database⁸ (downloaded on 2019/01/07). To make comparisons easier, the data were then filtered on the basis of depth (only entries between 200–1,000 m) and taxa (zooplankton only), and only quantitative records of biomass and abundance were included. After applying these filters, the dataset contained 89,308 entries. However, the filtered dataset did not include any records from the North Atlantic Ocean basin, despite significant sampling efforts in this area (Figure S3b).

The BioTIME database¹⁰ (downloaded on 2018/09/03) is a comprehensive collection of biodiversity time series comprising records of abundances and biomass for a range of marine and nonmarine species. The marine realm alone contains approximately 5,664,196 records, represented by 152 studies across six categories (benthic, fish, marine birds, marine invertebrates, marine mammals, and marine plants). However, extracting data on the basis of depth or elevation is challenging, as no depth information was available in the downloaded file. To address this, a file with depth information was obtained through personal communication with the authors and linked with the main query using the STUDY_ID as a common field. The resulting data were filtered for a depth range between 200 and 1,000 meters. However, some studies in the database lack depth information, resulting in significant data reduction. In addition, only two taxa (fish and marine invertebrates) were included in the current study, resulting in only 6,720 entries from three studies. Fish records (n = 719) were collected with 1.3 cm mesh and could not be classified as mesozooplankton or micronekton and were removed from the analysis, leaving 5,990 records for zooplankton (Figure S3b). Unfortunately, information on mesh size or net type was not available for these studies.

The Jellyfish Initiative (JeDI) Database²⁵ is a comprehensive global database dedicated to gelatinous zooplankton (Cnidaria, Ctenophora and Thaliacea), aimed at defining a global baseline of gelatinous zooplankton populations²⁶. The database consists of 476,000 quantitative, categorical, presence-absence, and presence-only records spanning three centuries (1790–2011), gathered from various published and unpublished sources. These records were collected globally, with the greatest concentration of data in the mid-latitudes of the Northern Hemisphere. After filtering for depths greater than 200 m and removing two density columns (density and density_integrated) due to a lack of identified units, the data were further filtered for nonzero quantitative density or biomass values (Figure S3b). This approach resulted in 8,413 entries remaining for analysis. Additionally, errors in units for mesh sizes reported in millimeters versus micrometers were corrected.

All data sources were compiled together in a single database for further standardization and cleanup. Depending on the date of collection/entry to the database, some taxonomic names can be entered as abbreviated names (e.g., S. gazellae), outdated names (e.g., Phyllopus helgae is unaccepted name for Nullosetigera helgae), or misspelled names (e.g., Conchoecinae instead of Conchoeciinae). To standardize the naming of different taxa, names were checked with the World Register of Marine Species⁷ (WoRMS), an authoritative classification and catalog of marine names. We used the worrms package to match the species names to their accepted WoRMS names²⁷. The original names were retained, and the standardized values were put into the taxa_standardized column along with AphiaID (aphia_id column). In addition, for each organism, taxonomic rank was specified. The exceptions were paraphyletic terms such as Gammaridea or Natantia, where precise taxonomic classification was not available, or when several taxa were listed together (e.g., copepods, chaetognaths or gelatinous zooplankton). In these cases, the rank ‘Group’ was assigned to taxa_standardized and rank columns. In cases where the community of organisms was given (e.g., Mesozooplankton and/or Micronekton), standardized name was assigned to ‘Zooplankton/ Micronekton.’ When organisms were recorded as “unidentified” or “other,” the standardized name were assigned to ‘Other taxa.” Records of terrestrial taxa (e.g., the genus Mesoniscus) were removed from the database.

Standardization for net types involved consolidating various net names, abbreviations, and descriptions into consistent categories. Similar or synonymous terms for the same net type, such as “Bongo,” “BN,” and “Bongo Net,” were grouped under a single standard name. Acronyms and abbreviations were aligned to their full forms where possible, and different variations of net descriptions were unified. For example, “IKMT” was standardized across variations like “Isaac-Kidd Midwater Trawl” and “Isaacs-Kidd midwater trawl.” Nets with the same function but different local variations were also combined, and generic terms like “Plankton Net” were clarified when possible. This process ensured consistency across different data sources and improved clarity for analysis. In addition, volumetric abundance values were standardized based on mesh size, tow type, and season, following the detailed procedure outlined in the Technical Validation section.

Data Records

The dataset is provided as a single Excel file (MMMD.xlsx) and is openly accessible via Zenodo²⁸ under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. Version (v1) of the dataset was peer reviewed. The file contains four tabs: (1) MMMD, which includes the main compiled dataset of mesopelagic mesoplankton and micronekton records drawn from various sources, detailing sampling location, depth, net type, volume filtered, abundance or biomass, and—where available—taxonomic group and calibration factors for mesh size, tow type, and season; (2) metadata_column_descriptions, providing definitions for each column in the main data table; (3) metadata_quant_units, describing the measurement units used across datasets; and (4) metadata_nets_abbreviation, listing verbatim and standardized net types to support consistency across sources. All fields and their descriptions are available in the accompanying file on Zenodo and are detailed in Table S1.

Technical Validation

Dataset summary

The Mesopelagic Mesozooplankton and Micronekton Database (MMMD) was compiled from all available literature (published and unpublished) containing depth-stratified information on mesopelagic species distributions and abundances. The database is in a single table with an accompanying table of column descriptions. The final database has ~270,000 entries and 46 columns collected from 258 different data sources, spanning the years 1880 to 2016 and 5,557 unique locations.

The 2D spatial distribution of the available records is extensive, covering all oceans (Fig. 1). The least sampling was performed in the center of ocean basins and the Indian sector of the Southern Ocean. Although the database has extensive coverage, only few records has information on the size of the organisms collected (Fig. 2). Relying solely on net mesh size information—when actual organism size is absent—introduces serious limitations for ecological analyses. Size is a fundamental trait that governs trophic interactions, metabolic rates, and contributions to carbon export²⁹.

Fig. 1.

Global distribution of records in the Mesopelagic Mesozooplankton and Micronekton Database colored by the number of the records. Note the log₁₀ transformation of the color scale. The map is in Eckert IV global equal area projection.

Fig. 2.

Number of missing biotic and abiotic information collected in the Mesopelagic Mesozooplankton and Micronekton Database.

The MMMD has 16 phyla (Table 2). Among these phyla, the phylum Arthropoda accounted for the majority, accounting for 70.5% of the records. The distributions of the other phyla were as follows: Cnidaria, 7.4%; Annelida, 6.5%; Chordata and Chaetognatha, 6.5%; Mollusca, 2.6%; Ctenophora, 0.2%; Echinodermata and Nemertea, 0.1%. The remaining phyla collectively contributed to less than 0.1% of the total records. The majority of the data are available at low taxonomic resolution: 55.5% of the data were recorded at the species and subspecies levels, and 12.5% were recorded at the genus level (Table 3). A large proportion of the entries were also community entries (i.e., Group; 5.1%). Out of 16 phyla, only 8 contained information at species level (Table S2). The number of species for each of the phyla is shown in Fig. 3.

Table 2.

Frequency of occurrence of various phyla in the database.

Phylum	Number of entries (% of all entries)
Arthropoda	263875 (70.5%)
Cnidaria	27781 (7.4%)
Annelida	24456 (6.5%)
Chordata	24369 (6.5%)
Chaetognatha	22521 (6.0%)
Mollusca	9871 (2.6%)
Ctenophora	641 (0.2%)
Echinodermata	401 (0.1%)
Nemertea	266 (0.1%)
Phoronida	170 (<0.1%)
Bryozoa	101 (<0.1%)
Nematoda	24 (<0.1%)
Porifera	14 (<0.1%)
Brachiopoda	12 (<0.1%)
Hemichordata	4 (<0.1%)
Platyhelminthes	4 (<0.1%)

Table 3.

Taxonomic resolution.

Taxonomic rank	Number of entries (% of all entries)
Kingdom	2 (<0.1%)
Gigaclass	15 (<0.1%)
Phylum	12466 (3.8%)
Subphylum	1521 (0.5%)
Infraphylum	4 (<0.1%)
Superclass	533 (0.2%)
Class	34686 (10.7%)
Subclass	1380 (0.4%)
Superorder	236 (0.1%)
Order	25520 (7.8%)
Suborder	1521 (0.5%)
Infraorder	329 (0.1%)
Superfamily	270 (0.1%)
Family	8751 (2.7%)
Subfamily	25 (<0.1%)
Genus	40791 (12.5%)
Species	157980 (48.6%)
Subspecies	22388 (6.9%)
Group	16686 (5.1%)

Frequency of occurrence of different taxonomic units in the database. The rank zooplankton or micronekton is given for entries where density/biomass is given for the entire community of organisms. A group is assigned when species are reported in groups or when two different ranks are combined.

Fig. 3.

Phylogeny of life contained within the Mesopelagic Mesozooplankton and Micronekton Database for all taxa recorded at the species level. Here, the diversity of organisms is based on their taxonomic classification (phylum, class, order, family). The gray bars reflect the log₁₀-transformed number of species recorded in each family. The N/S in the diagram (phylum Cnidaria) represents an unidentified family for Rhabdoon reesi. All silhouettes were obtained from Phylopic (http://phylopic.org).

Fifty-four different sampling methods were recorded in the MMMD (Table 4). Juday net was the most common sampling gear in the database (23.6% of entries), followed by MOCNESS Plankton net, and RMT8 (16.1%, 5.8% and 5.4% of entries, respectively). Three types of samples were collected: vertical (45.8% of entries), oblique (26.7% of entries), horizontal (5.9% of entries) or combinations of the three (<0.3%). No information was recorded for 21% of the entries. Samples with a variety of mesh sizes ranging from 30 µm to 2.2 cm were collected (Fig. 4a,b). In total, 63.2% of the entries had no record of the sampling time (Fig. 2). Only 14.8% and 17.6% of the records could be identified as day- and night-time collections, respectively (Fig. 4d). Most of the records were collected after 1960 (Fig. 4c), however quite a few records were done between 1880 and 1920. Finally, 3.4% of the records contained pooled abundance/biomass estimates from both day and night. The most abundant quantitative data were density (ind.∙m⁻³) and areal abundance (ind.∙m⁻²), which accounted for 48.5% and 21.3% of all records, respectively (Table 5).

Table 4.

Top 10 sampling methods and associated mesh size and net mouth areas in the Mesopelagic Mesozooplankton and Micronekton Database.

Sampling Method	Number of entries (% of all entries)	Possible mesh sizes, µm	Possible mouth areas, m2
BIONESS (Bedford Institute of Oceanography Net and Environmental Sampling System)	7870 (2.4%)	200–500	1, 2.1
Bogorov – Ross net	17309 (5.3%)	112–530	1
Bongo Plankton Sampler	7478 (2.3%)	200–202	—
DzhOM (Oceanic modification of the Juday) net	8284 (2.5%)	160	0.5
Juday net	76640 (23.6%)	112–569	01.0.5
MOCNESS (Multiple Opening and Closing Net System)	52219 (16.1%)	149–335, 505, 3000	0.79,1,10
MPN - HYDROBIOS (Multiple Plankton Net)	10691 (3.3%)	55–300	0.125, 0.25, 1
MTD (Motoda horizontal closing nets)	10288 (3.2%)	350	—
Multiple Plankton Sampler	7575 (2.3%)	180–202	0.49
Multiple sampling techniques	11129 (3.4%)	various	various
North Pacific Standard Plankton Net	10377 (3.2%)	90–350	0.2–1
Plankton net	18947 (5.8%)	30–750	0.2–5.3
RMT1 + 8 (rectangular midwater trawl system)	5000 (1.5%)	300–330, 4500	1, 8
RMT8	17494 (5.4%)	300, 4500, 5000	3–11
Submersible observations	9618 (3%)	—
Vertical Multiple Plankton Sampler	8630 (2.7%)	60–330	0.2, 1
Other catch methods	45555 (14%)	—

Fig. 4.

Characteristics of Mesozooplankton and Micronekton Database. Distribution of mesh sizes for (a) Meso- and (b) Macro- zooplankton. Distribution of records based on (c) year and (d) time of sampling. Note that Bathypelagial was entered as time for sampling in the data obtained from PANGAEA.

Table 5.

Frequency of occurrence of data recorded in different abundance/biomass units in the mesopelagic database.

Biomass/Abundance Units	Number of entries (% of all entries)
Abundance (ind.∙m−3)	117,026 (43.3%)
Abundance (ind. ∙m−2)	62,927 (23.4%)
Catch rate (ind.∙haul∙h−1)	38,738 (14%)
Abundance (# ind.)	13,786 (5.0%)
Biomass (g∙m−3)	5,453 (2.0%)
Biomass (g wet Wt. ∙m−3)	4,682 1.7%)
% Abundance of total # ind.	4,622 (1.7%)
Biomass (mg C ∙m−3)	4,078 (1.5%)
Biomass (g dry Wt. ∙m−3)	3,773 (1.4%)
Biovolume (ml∙m−3)	3,503 (1.3%)
Biomass composition (%)	2,409 (0.9%)
Biomass (mg C ∙m−2)	1,854 (0.7%)
Total number of species (#)	1,805 (0.7%)
Biomass (g dry Wt.∙m−2)	1,673 (0.6%)
Biomass (g wet Wt.∙m−2)	1,124 (0.4%)
Catch rate (ind.∙hour−1)	674 (0.2%)
Catch rate (kg∙h−1)	564 (0.2%)
Biomass (µg dry Wt.∙ind−1)	240 (0.1%)
Biomass (g∙m−2)	204 (0.1%)
Displacement volume (ml∙m−2)	109 (<0.1%)
Biomass (g dry Wt.∙ind.−1)	51 (<0.1%)
Biomass (µg C∙ind.−1)	51 (<0.1%)
Catch rate (ml∙haul−1)	14 (<0.1%)
Biovolume (ind. ∙ml−1)	3 (<0.1%)
No Biomass/abundance value	6969 (2.5%)

Abbreviations: ind. - individuals; Wt. - weight; # - number; C- carbon.

Abundance calibration

Although extensive quantitative information exists on the distributions of various mesopelagic mesozooplankton and micronekton species, regional comparisons are challenging due to differences in data collection times, methods, and sampling methods. Additional factors, such as net mouth area and organism type, further complicate these comparisons. This work aims to standardize data with adjusted density values. The subset of the MMMD used comprises 112,000 entries (42.3% of all records) for densities, including 28,120 entries with zero densities, which could bias linear models (see Supplemental Information: Abundance Standardization for more details). To address this, a logarithmic transformation was applied, adding a constant (e.g., 0.001) to each density value. However, to avoid biasing lower density estimates during calibration, zero entries were excluded, allowing logarithm calculations without constant addition. This method prepares the database for accurate analysis by mitigating biases from zero densities. Before analysis, several columns were standardized to reduce the number of groups. The MMMD, with 7,858 unique tows, was classified into mesozooplankton and micronekton based on mesh size, with a cutoff of 1,000 µm. Mesozooplankton were categorized into six mesh classes (≤100 µm, 100–200 µm, 200–250 µm, 250–350 µm, 350–500 µm, and 500–1,000 µm), whereas micronekton were categorized into three mesh classes (1,000–3,000 µm, 4,500–5,000 µm, and >5,000 µm). Owing to fewer micronekton records, standardization focused on mesozooplankton, using the 225 µm mesh class as most frequently used for standardization.

In the taxonomic standardization process, several phyla, such as Porifera and Platyhelminthes, were excluded because of insufficient data. The phyla Ctenophora and Echinodermata had low counts and were analyzed at the phylum level. The remaining phyla, including Annelida, Mollusca, Arthropoda, Cnidaria, and Chordata, were further divided into taxonomic classes. Within Annelida, most records were from the class Polychaeta, which remained at the class level because of unspecified taxonomic orders. Mollusca records were predominantly from the class Gastropoda, with other classes, such as Cephalopoda and Bivalvia, also included. Arthropoda includes the classes Copepoda, Malacostraca, and Ostracoda, with Malacostraca further divided into orders such as Euphausiacea and Amphipoda. Cnidaria records were primarily from the class Hydrozoa, with specific orders such as Siphonophorae included. Chordata was represented by classes Teleostei, Appendicularia, and Thaliacea. A total of 24 taxa were selected for calibration, accounting for 90% of the records with nonzero mesozooplankton densities. The records were stratified spatially and seasonally to reduce variability.

Calibration involves sequential steps to standardize the mesh size, tow type, and season and uses linear regression models to adjust the densities to a reference mesh class, tow type, and season. This approach effectively corrected for spatial and methodological variability in the data. To illustrate the calibration process, a case of order Euphausiacea is presented (Fig. 5). Euphausiacea was chosen as an example of a taxon with many records in the MMMD. In addition, this taxon was caught with different mesh sizes and tows and during different seasons. The distribution of raw densities (log-transformed) was variable. However, after standardization, the distribution of log-transformed densities looks more symmetrical and is centered around the same mean.

Fig. 5.

Example of the calibration procedure for order Euphausiacea colored by different mesh sizes (Mesh classes of 50, 150, 225, 300, 400 and 750 µm). (a) Raw densities and (b) densities after correction for different mesh classes; (c) densities from step B corrected for the tow type; (d) densities from step C corrected for the season. Note the log₁₀ transformation on the $x$ axis.

Usage Notes

This database is a composite source of all the available data from various sources, so care is needed when these data are used and interpreted because of the different sampling methods used. Although every effort was made to gather all available data, the database is likely incomplete, as many datasets are not published. Therefore, it is envisaged that the database will encourage scientists around the world to submit published but not publicly available datasets.

Author contributions

E.A.P., I.M., A.L., Y.E. and T.S. collected data; Y.E. performed data standardization and compilation of all the resources; Y.E. and E.P. wrote the paper; all authors edited the paper. All authors have read and agreed to the published version of the manuscript.

Code availability

The code is publicly available in the following GitHub repository: https://github.com/yuliaUU/mmmd.git. The repository encompasses the R code that was used to perform the calibration of abundance values and generate the images in this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41597-025-05638-w.