Agricultural practices for cleaner water: A global synthesis of meta-evidence

Agricultural intensification has pushed nitrogen and phosphorus cycles beyond safe planetary limits, with agriculture now a primary driver of at least two Earth system boundaries being breached. Technological fixes alone are unlikely to resolve this: efficiency gains tend to lower production costs and drive further expansion (known as Jevon’s Paradox), intensifying environmental pressure rather than relieving it. Reducing agricultural water pollution will therefore require a shift toward production systems that work with ecological processes rather than against them. Despite growing interest in agroecological and regenerative approaches, evidence on their effectiveness for improving water quality remains fragmented across practices, regions, and study designs. This review of reviews synthesises global evidence from 67 multi-continental meta-analyses on how farm practices affect nutrient, pesticide and sediment losses. Non legume cover crops cut nitrate leaching by about 50 to 70 percent and sediment by 50 to 60 percent. Residue retention and organic amendments reduced runoff nutrients by 25 to 50 percent and sediment by up to 75 percent. Reduced tillage lowered sediment losses by about 60 percent but often increased nitrate leaching, highlighting the importance of complementary fertiliser management. Optimising fertiliser rates, timing, and formulation, alongside precision irrigation, reduced nitrogen losses by 20 to 70 percent. At the landscape scale, vegetated buffers, agroforestry, and wetlands typically removed 25 to 90 percent of nutrients and 40 to 95 percent of sediment. Improving soil cover, structure, and hydrological function can sharply reduce pollution without sacrificing yields, showing that poor water quality stems from management, not inevitability. However, important evidence gaps remain: pesticide transport lacks any global synthesis despite residues being among the most pervasive agricultural contaminants, quantitative evidence on adaptive grazing management remains limited, and most underlying studies are concentrated in temperate croplands of North America, Europe, and China, limiting confidence in applying these findings to tropical, arid, or smallholder systems.