Abstract
Water quality impairment by elevated sediment loss is a pervasive problem for global water resources. Sediment management targets identify exceedance or the sediment loss 'gap' requiring mitigation. In the UK, palaeo-limnological reconstruction of sediment loss during the 100-150 years pre-dating the post-World War II intensification of agriculture, has identified management targets (0.20-0.35 t ha(-1) yr(-1)) representing 'modern background sediment delivery to rivers'. To assess exceedance on land for grazing ruminant farming, an integrated approach combined new mechanistic evidence from a heavily-instrumented experimental farm platform and a scaling out framework of modelled commercial grazing ruminant farms in similar environmental settings. Monitoring (2012-2016) on the instrumented farm platform returned sediment loss ranges of 0.11-0.14 t ha(-1) yr(-1) and 0.21-0.25 t ha(-1) yr(-1) on permanent pasture, compared with between 0.19-0.23 t ha(-1) yr(-1) and 0.43-0.50 t ha(-1) yr(-1)and 0.10-0.13 t ha(-1) yr(-1)and 0.25-0.30 t ha(-1) yr(-1) on pasture with scheduled plough and reseeds. Excess sediment loss existed on all three farm platform treatments but was more extensive on the two treatments with scheduled plough and reseeds. Excessive sediment loss from land used by grazing ruminant farming more strategically across England, was estimated to be up to >0.2 t ha(-1) yr(-1). Modelled scenarios of alternative farming futures, based on either increased uptake of interventions typically recommended by visual farm audits, or interventions selected using new mechanistic understanding for sediment loss from the instrumented farm platform, returned minimum sediment loss reductions. On the farm platform these were 2.1 % (up to 0.007 t ha(-1) yr(-1)) and 5.1 % (up to 0.018 t ha(-1) yr-1). More strategically, these were up to 2.8 % (0.014 t ha(-1) yr(-1)) and 4.1 % (0.023 t ha(-1) yr(-1)). Conventional on-farm measures will therefore not fully mitigate the sediment loss gap, meaning that more severe land cover change is required.