Lake Erie tributary nutrient trend evaluation: Normalizing concentrations and loads to reduce flow variability

• Published in: Ecological indicators 2021-06, Vol.125, p.107601, Article 107601
• Main Authors: Rowland, Freya E., Stow, Craig A., Johnson, Laura T., Hirsch, Robert M.
• Format: Article
• Language: English
• Subjects: Long-term data; Monitoring; Nutrients; Statistics; Streams; Time-series analysis
• ISSN: 1470-160X; 1872-7034
• DOI: 10.1016/j.ecolind.2021.107601

[Display omitted] •Assessing changes over time in stream nutrients is hard because of flow variability.•Flow normalizing made patterns more perceptible.•Flow-normalization removed the noise arising from annual precipitation.•Flow normalizing makes tributary nutrient trend evaluation more straightforward. Establishing tributary load (i.e., the mass exported over a period of time) targets to reduce anthropogenic nutrient inputs to receiving waters — and thus eutrophication — is a common mitigation strategy in freshwater and coastal ecosystems. However, detecting and quantifying trends can be difficult because annual precipitation strongly influences tributary flow (e.g., average daily stream discharge). This may obscure trends as wet years tend to produce high tributary loads despite management activities to reduce nutrient export, and dry years typically generate low loads, even without management of nutrients. Furthermore, flow and nutrient concentrations are often correlated. Earlier efforts to reduce the effect of flow variability on tributary nutrient assessment were limited by computational and methodological constraints, until the weighted regressions on time, discharge, and season (WRTDS) method was introduced in 2010. Here we use WRTDS to assess nutrient concentration and load changes from 1982 to 2018 in three tributaries to the western basin of Lake Erie, of the Laurentian Great Lakes. Generally, trends revealed by flow-normalization do not contradict those of non-normalized metrics; however flow-normalization made the patterns more perceptible than in non-normalized metrics and reduced the influence of a particularly wet or dry period at the end of records on long-term trend analysis. We demonstrate that using WRTDS for flow-normalization removed the noise arising from annual precipitation variability and makes tributary nutrient trend evaluation more straightforward.