A datamining approach to identifying spatial patterns of phosphorus forms in the Stormwater Treatment Areas in the Everglades, US

Abstract

The Everglades ecosystem in Florida, USA, is naturally phosphorus (P) limited, and faces threats of ecosystem change and associated losses to habitat, biodiversity, and ecosystem function if subjected to high inflows of P and other nutrients. In addition to changes in historic hydropattern, upstream agriculture (sugar cane, vegetable, citrus) and urbanization has placed the Everglades at risk due to nutrient-rich runoff. In response to this threat, the Stormwater Treatment Areas (STAs) were constructed along the northern boundary of the Everglades as engineered ecological systems designed to retain P from water flowing into the Everglades. This research investigated data collected over a period from 2002 to 2014 from the interior of the STAs using data mining and analysis techniques including (a) exploratory methods such as Principal Component Analysis to test for patterns and groupings in the data, and (b) modelling approaches to test for predictive relationships between environmental variables. The purpose of this research was to reveal and compare spatial trends and relationships between environmental variables across the various treatment cells, flow-ways, and STAs. Common spatial patterns and their drivers indicated that the flow-ways do not function along simple linear gradients; instead forming zonal patterns of P distribution that may increasingly align with the predominant flow path over time. Findings also indicate that the primary drivers of the spatial distribution of P in many of these systems relate to soil characteristics. The results suggest that coupled cycles may be a key component of these systems; i.e. the movement and transformation of P is coupled to that of nitrogen (N).