Abstract The adverse impacts of agriculture on water quality have been well documented and the practical measures required to mitigate excessive losses of nutrients and fine sediment have been demonstrated to be effective at plot to field scales. Evaluating the long-term, cumulative impact of measures is challenging, however, and the evidence for tangible water quality benefits at a catchment scale is limited. The objective of this study was to evaluate the effectiveness of the Catchment Sensitive Farming (CSF) partnership, a voluntary pollution reduction scheme in England that between 2006 and 2018 provided advice to 19,776 farm holdings covering 34 % of the total farmed area in England. Water quality data from 110 river monitoring sites in catchments ranging from 11 km2 to 2276 km2 was analysed using statistical regression models to compare pollutant concentration trends among sites that varied in the amount and timing of upstream pollution reduction activity. Additional covariates were included to control for potentially confounding changes in pollution pressure caused by: (i) background changes in cropping patterns and livestock numbers, (ii) the uptake of land management measures delivered via other policy mechanisms, and (iii) seasonal and weather-driven variation in water quality. After delivering advice to farmers, there was a delay of at least two years before mitigation measures elicited a detectable improvement in suspended solids (SS) and orthophosphate (OP) at downstream monitoring sites. Across all sites, the percentage reduction in SS concentration was 28 % of the modelled percentage reduction in fine sediment load. The strength of response was site-specific and changes in SS concentration varied from +12 % to -33 %. In absolute terms, CSF was estimated to have reduced the mean SS concentration at sites in target catchments by an average of 4.4 %. Mean OP concentrations in target catchments decreased by an average of 13.0 % (95 % confidence interval 3.7–22.4%) but the overall strength of the water quality response to unit reductions in phosphorus loading was uncertain. CSF measures reduced nitrogen loading by 5 % in target catchments, but there was no evidence that mean concentrations of total oxidised nitrogen (TON) decreased, probably due to the complex and site-specific time lags involved. The combination of pollution load modelling and statistical water quality modelling used in this study provides a framework for testing and quantifying the effectiveness of land management interventions, individually and in combination, but also highlights the importance of establishing a clear counterfactual and accounting for spatially-variable, lagged responses.

中文翻译：

---

水质受益于以建议为导向的减少英格兰农业水污染的方法

摘要 农业对水质的不利影响已被充分证明，减轻营养物质和细粒沉积物过度损失所需的实际措施已被证明在田间尺度上是有效的。然而，评估措施的长期累积影响具有挑战性，并且在流域范围内有形水质效益的证据是有限的。本研究的目的是评估集水区敏感农业 (CSF) 合作伙伴关系的有效性，这是英格兰的一项自愿减少污染计划，该计划在 2006 年至 2018 年间为 19,776 个农场提供了建议，覆盖了英格兰总养殖面积的 34%。使用统计回归模型分析了来自 11 平方公里至 2276 平方公里集水区 110 个河流监测站点的水质数据，以比较上游污染减少活动数量和时间不同的站点之间的污染物浓度趋势。还包括其他协变量，以控制由以下因素引起的污染压力的潜在混杂变化：(i) 种植模式和牲畜数量的背景变化，(ii) 通过其他政策机制实施的土地管理措施，以及 (iii) 季节性和天气驱动的水质变化。在向农民提供建议后，在缓解措施导致下游监测点悬浮固体 (SS) 和正磷酸盐 (OP) 出现明显改善之前，至少延迟了两年。在所有站点中，SS 浓度的减少百分比是细粒沉积物负荷模拟百分比减少的 28%。响应强度是特定于站点的，SS 浓度的变化从 +12% 到 -33% 不等。就绝对值而言，据估计 CSF 已将目标流域中站点的平均 SS 浓度平均降低了 4.4%。目标流域中的平均 OP 浓度平均下降了 13.0%（95% 置信区间为 3.7-22.4%），但水质对单位磷负荷减少的总体反应强度尚不确定。CSF 措施将目标流域的氮负荷降低了 5%，但没有证据表明总氧化氮 (TON) 的平均浓度降低，这可能是由于所涉及的复杂和特定地点的时间滞后。