Globally, significant societal resources are devoted to mitigating negative effects of eutrophication from excessive phosphorus (P) and nitrogen (N) loading. Potential effectiveness of mitigation measures and possible confounding factors are often assessed using studies conducted in headwater catchments. However, success is often evaluated based on trends in river mouth water chemistry. It is not clear how transferrable insights from headwater catchments are to larger rivers. Here, relationships between P and suspended solids (SS) identified in small agricultural headwater catchments were applied to 30 larger, mixed land use catchments draining into Mälaren, a Swedish great lake. Relationships identified in headwater streams between SS concentration, catchment agricultural land percentage and arable land clay content were corroborated for the larger catchments (R2 = 0.59, p-value<0.001. The same was true for connections between SS and particulate P (R2 = 0.74, p-value<0.001). This study highlights the importance of agricultural land, clay content and SS for P transport, on both smaller headwater as well as larger catchment scales, supporting the use of headwater findings on larger, management relevant scales. Consequently, these relationships should be used to target mitigation measures to reduce SS and P losses. To explore the effectiveness of mitigation measures on water quality, we assessed long-term (20 year) trends in tributary water quality and compared these trends to the amount of mitigation measures implemented in the catchment. Overall improving trends were detected using regional Mann Kendall tests, but few decreasing trends in nutrient concentrations were found for individual sites using Generalized Additive Models (GAM). The lack of significant trends and identifiable connections to amount of mitigation measures implemented could be due to several reasons, e.g. insufficient time for recently implemented measures to have an effect, ongoing release of legacy P as well as low areal coverage and poor spatial placement of implemented measures. In addition, trend detection requires large amounts of data and the results should be carefully interpreted and communicated.
Read full abstract