Global staple-food trade dataset estimates biodiversity loss across 157 countries, 1995–2022

A research dataset and network analysis estimated biodiversity loss embodied in international trade of wheat, soybeans, rice and maize across 157 countries from 1995 through 2022. Published on 14 January 2026, the resource is intended to document how commodity flows and country relationships changed over the 28-year period. The estimates concern modeled species-richness loss associated with land-use change, rather than direct measurements of biodiversity decline at trade destinations.

The study linked bilateral trade volumes with biodiversity-loss intensity factors to create a weighted, directed network. Its 157 nodes represent countries, while directed edges represent bilateral trade flows. Edge weights indicate the estimated quantity of biodiversity loss embodied in the traded commodity and transfer from an exporting country to an importing country.

Trade data for the four staple crops came from UN Comtrade and were classified using six-digit Harmonised System codes. Annual production per unit area, or yield, was obtained from FAOSTAT. The researchers used Potential Disappeared Fraction factors from Cabernard and colleagues’ 2024 study to represent the proportion of species richness lost through land-use change.

Each biodiversity-loss intensity factor was divided by the corresponding annual yield to produce a quality-adjusted factor intended to improve comparability across regions and years. For every directed trade flow, the adjusted factor was multiplied by the annual trade volume of the relevant commodity. The resulting value became the weight of that network edge. The researchers then calculated average degree, graph density, clustering coefficient, average path length, number of communities, modularity and the standard deviation of biodiversity-loss values across countries.

The number of recorded trade edges increased from 1,895 in 1995 to 91,414 in 2022, representing a roughly 47-fold rise. This measured expansion indicates that the dataset contains a much denser set of country-to-country relationships at the end of the period than at the beginning. It does not by itself establish that every recorded relationship produced additional biodiversity loss, because edge weights also depend on commodity volumes and the underlying loss-intensity factors.

Estimated embodied biodiversity loss increased substantially over the full period. Soybeans recorded the fastest growth among the four commodities, with their estimated loss rising by a factor of 6.35, from 1.61 × 10^-6 PDF in 1995 to 1.02 × 10^-5 PDF in 2022. The corresponding increases were 3.29-fold for rice, 3.07-fold for maize and 1.69-fold for wheat. Wheat historically represented the largest share of total embodied biodiversity loss in the network, but soybeans overtook it in 2022 to become the largest contributor.

Countries with substantial agricultural trade showed the greatest average embodied biodiversity loss associated with imports and exports over the study period. The analysis highlights the United States, Brazil, China, Australia and Argentina in this respect. It also identifies bilateral routes connecting major agricultural producers and consumers as important pathways for transferring embodied biodiversity loss across borders. These results describe patterns in the constructed trade-and-biodiversity network rather than measured changes in local species richness.

The researchers excluded trade records for which a relationship was known but the traded volume was missing, distinguishing these records from genuine zero-trade relationships. They described the missing volume as a minor proportion and reported no systematic concentration in particular years or countries. The authors state that excluding these records should have a negligible effect on the network’s macro-structure and main conclusions, although the released data do not include embodied-loss estimates for every potential trade relationship.

The estimates have several bounded uncertainties. Biodiversity-loss intensity factors are outputs of complex models and assumptions rather than absolute measurements, and integrating FAOSTAT yield data can propagate errors from agricultural production statistics. Potential Disappeared Fraction addresses species-richness loss caused by land-use change but does not capture every dimension of biodiversity decline. The authors suggest sensitivity analysis using a multi-regional input-output or similar scenario framework to test network dynamics and country rankings against alternative factor bounds and trade-volume assumptions.

The resource is stored in CSV format, and the custom research code is publicly available on GitHub. The dataset was created in March 2025, before the 2026 publication. The authors describe it as the first comprehensive and dynamic network dataset focused on embodied biodiversity loss in global staple-food trade and position it as infrastructure for research in ecology, economics, geography and policy. Its potential uses include evaluating environmental implications of trade relationships and informing work on biodiversity conservation and sustainable food trade, but the published resource does not establish the effects of any particular policy.

Subject of Research: Network-based estimates of biodiversity loss embodied in global staple-food trade

Article Title: Dataset of Biodiversity Loss in Global Staple Food Trade, 1995-2022

Article References: Huang, Z., He, Z., & Xing, Z. (2026). Dataset of Biodiversity Loss in Global Staple Food Trade, 1995-2022. One Ecosystem, 11, Article e159884. https://doi.org/10.3897/oneeco.11.e159884

Image Credits: AI Generated

DOI: 10.3897/oneeco.11.e159884

Keywords: biodiversity loss, staple food trade, global trade, food security, land-use change, species richness, network analysis, UN Comtrade, FAOSTAT, wheat, soybeans, rice