A phase space approach to understanding the efficacy of agricultural BMP adoption on water quality under a range of socio-ecological scenarios using a watershed scale, feed-forward integrated assessment model

The health of waters across the country and around the world is compromised by excessive availability of nutrients, particularly phosphorus and nitrogen, leading to eutrophication, toxic algae blooms, and hypoxic dead zones. To address the problem, a suite of best management practices (BMPs) is typically recommended to control the mobility and downstream availability of nutrients thereby limiting the likelihood of developing harmful eutrophic conditions. This approach has had mixed results in practice, and assumed efficacy rates are further complicated by changing climatic conditions and extreme events. The present study advances an integrated assessment model that utilizes a feed-forward modeling approach at the watershed scale to assess the range of efficacy of implementing agricultural BMPs on water quality under various scenarios using a phase-space analysis across multiple social-ecological dimensions. Agriculture in particular is a significant contributor to global aquatic nutrient pollution. According to the most recent US Environmental Protection Agency (EPA) analysis, in the Missisquoi watershed of the Lake Champlain basin, the focal social-ecological system for this study, agriculture was estimated to contribute 42% of the phosphorus (P) pollution between 2000-2010. As a result of that study, US EPA recommended an 80% reduction in P pollution from the agricultural sector to meet water quality targets by 2035. The suite of scenarios considered in this study therefore encompasses social and environmental dimensions within the following three categories: (i) likelihood of BMP adoption, (ii) P reduction efficacy of the BMPs adopted, and (iii) variation within the hydrologic cycle. Results are presented that explore the phase space of harmful algal bloom (HAB) conditions under these scenarios given various model assumptions. The results of the present study will further inform decision-making at the watershed scale and demonstrate the capability of an integrated assessment modeling approach to inform watershed scale management.