A multilateral development bank financed a feasibility study to rehabilitate irrigation infrastructure across Mongolia, part of a wider push on food security and climate resilience. AtkinsRéalis, delivering the project's Environmental and Social Impact Appraisal, needed to know which fish were at risk of entrainment at the schemes' offtake structures — in river systems with almost no existing biodiversity data.

Mongolia's rivers run frozen for much of the year and reappear each summer as fast-changing, remote channels with little historical survey coverage. Electric fishing, the conventional method, is hard to deploy safely at this scale and remoteness, and the bank's environmental requirements — aligned with TNFD, GRI, and ISSB disclosure expectations — demanded a credible fish baseline before construction could proceed.

AtkinsRéalis built aquatic eDNA into its existing field survey programme, using NatureMetrics' kits to collect and filter up to 2 litres of water at each scheme's offtake point and at 1 km intervals up- and downstream. Samples were shipped to NatureMetrics' UK lab and processed using fish metabarcoding.

All 21 samples made it back from Mongolia to the UK, and 81% returned reportable fish eDNA results — identifying 19 taxa in total: 12 to species level, 5 to genus, 2 to family. Three of those were regionally vulnerable or near-threatened: Arctic grayling (Thymallus arcticus), the endemic Mongolian grayling (Thymallus brevirostris), and lenok (Brachymystax lenok). Just as tellingly, eDNA picked up 4 species the electric fishing surveys hadn't recorded at all.

The results gave AtkinsRéalis and the bank the evidence to classify each scheme's fish risk and design mitigation — from reducing entrainment and impingement to improving fish passage — while strengthening the project's compliance case against TNFD, GRI, and ISSB. The ESIA has already recommended repeat surveys to capture seasonal variation, and as Mongolia's first application of eDNA, the project positions the approach as a template for MDB-financed infrastructure work in other remote, data-poor regions.