New R Package Turns Six Decades of Atlantic Trawl Survey Data into One-Click Science

Hidden beneath the waves of the Northeast Atlantic lies one of the most valuable long-term ecological datasets on Earth: more than 60 years of standardized bottom-trawl surveys, documenting the rise and fall of fish populations from the Baltic to the Bay of Biscay. Now, a team of Danish researchers has released a tool that promises to transform how scientists tap into this treasure trove, compressing thousands of lines of bespoke computer code into just a handful of commands.

The tool is called DATRASextra, an open-source package for the R statistical programming language, described in the journal SoftwareX by Tobias K. Mildenberger, Federico Maioli, and Casper W. Berg of the National Institute of Aquatic Resources at the Technical University of Denmark. It provides an end-to-end workflow for the International Council for the Exploration of the Sea’s Database of Trawl Surveys, known as DATRAS, a central repository that currently compiles information from 28 surveys spanning more than six decades of ocean monitoring.

The scale of the database is staggering. DATRAS contains records from approximately 142,000 individual trawl hauls, around 2,000 distinct taxonomic identifiers, and more than 20 million biological records distributed across multiple linked tables. Some survey time series stretch back to 1965, capturing everything from the abundance of commercially vital cod and herring stocks to the slow northward creep of warm-water species as the climate changes. These data underpin stock assessments for most of the 271 fish stocks managed in the region, as well as biodiversity monitoring, ecosystem-based management, and large-scale ecological syntheses.

Yet despite decades of standardization efforts, actually working with DATRAS data has long been a technical ordeal. The information is stored in relational tables: haul-level records describing where and how each trawl was towed, length-based records counting individuals of each species in each size class, and biological records detailing the age, weight, and maturity of sampled fish. Researchers traditionally had to download and merge these tables themselves, apply survey-specific filtering rules, implement quality-control checks, and construct analysis-ready datasets through extensive project-specific scripting. The result, the authors argue, is a patchwork of workflows that are difficult to reproduce, compare, and maintain across studies and institutions.

DATRASextra tackles this problem with a modular, workflow-oriented architecture. Data can be imported directly from the ICES web services or from locally archived exchange files, and are organized into a standardized object containing the three core record types. From there, a sequence of interoperable functions handles cleaning, enrichment, and quality control before exporting analysis-ready products. The package calculates swept-area estimates from haul metadata, derives numbers-at-length and biomass-at-length observations, aggregates catches into custom length classes, and computes stratified mean abundance and biomass indices, all while carrying hydrographic measurements such as surface and bottom temperature and salinity through the pipeline as environmental covariates.

A key design principle is transparency. The package’s default cleaning routine retains only hauls flagged as usable and records with complete standard-species recording, and harmonizes taxonomy so that species that cannot be reliably identified are consistently aggregated to genus. Every filter is exposed as an adjustable argument, and additional screening steps, such as outlier detection using rule-based bounds and percentile thresholds, are deliberately opt-in, so no observation is discarded without the user explicitly requesting it. The package even reports whether archived data still match the state in which they were downloaded, a crucial safeguard given that ICES revises historical records over time.

The practical payoff is dramatic. The Northeast Atlantic component of FishGlob, a major international initiative integrating bottom-trawl survey data across regions to study global patterns in marine fish communities, is currently generated in the public FishGlob repository by roughly 2,050 lines of survey-specific R code. In DATRASextra, the equivalent workflow consists of five function calls, or approximately ten lines of code including the data download. The package also includes a dedicated vignette reproducing that FishGlob component, which comprises around 45 percent of the surveys included in the global dataset.

The illustrative examples in the paper showcase the range of the tool. A minimal workflow maps spatial and temporal variation in Atlantic herring catch rates from the North Sea International Bottom Trawl Survey, revealing aggregations of high catch rates in waters between Denmark and Sweden in recent years. A second example derives length-structured indices for Northeast Atlantic wolffish, splitting the population into juvenile and adult groups using the species’ length at 50 percent maturity and revealing declining stratified abundance and biomass for both groups between 1983 and 2024. A third demonstrates cross-survey integration, producing maps of species richness across the entire DATRAS database for 2015 to 2020.

The package does not exist in a vacuum. It builds deliberately on earlier infrastructure: the icesDatras package provides a client for the ICES web services but stops at data retrieval, while the DATRAS package introduced the relational data object on which DATRASextra is built and supplies age-length key and growth-model tools that DATRASextra deliberately does not duplicate. Outside the ICES area, the surveyjoin package offers a comparable standardized database for Northeast Pacific surveys run by the United States and Canada, but its harmonization happens centrally during database assembly rather than under user control. The authors position DATRASextra as complementary, filling the gap between raw access and finished analysis for the world’s largest regional trawl survey database.

Adoption is already underway within the ICES expert community, with the package being used to investigate shifts in fish distribution and essential fish habitats, and to assess the impact of warming on growth rates and fishery yields. Released under the GNU General Public License with source code and documentation publicly available on GitHub, DATRASextra is actively developed, with planned extensions including gear standardization workflows, expanded life-history summaries, and routines for matching hauls to external gridded environmental reanalysis products. As marine survey datasets grow in size and importance for ecosystem-based management, tools that make their processing efficient, transparent, and reproducible are becoming indispensable, and this one lowers the barrier for anyone hoping to read six decades of ocean history written in fish.

Subject of Research: An R package for reproducible processing of ICES DATRAS bottom-trawl survey data

Article Title: DATRASextra: An R package for streamlined workflows with ICES DATRAS bottom-trawl survey data

Article References: Mildenberger, T. K., Maioli, F., & Berg, C. W. (2026). DATRASextra: An R package for streamlined workflows with ICES DATRAS bottom-trawl survey data. SoftwareX, 36, Article 103108. https://doi.org/10.1016/j.softx.2026.103108

Image Credits: AI Generated

DOI: 10.1016/j.softx.2026.103108

Keywords: DATRASextra, R package, bottom-trawl surveys, ICES DATRAS, fisheries stock assessment, biodiversity monitoring, FishGlob, reproducible workflows, marine ecology, Northeast Atlantic, open-source software, survey data