Anomalous subsurface phosphorus transport from field to stream in a tile drained landscape: Tracer experiment and models

In agricultural areas with poorly drained soils, subsurface tile drains are commonly installed to improve drainage but also serve as conduits that deliver excess nutrients to adjacent streams. Our goal was to understand the transport of phosphorus (P) along these flow paths by applying a novel mixture of tracers (including 866 g of conservative chloride (Cl), 3.4 g of potassium phosphate, and approximately 3.6x10¹¹ fluorescent micrometer-sized particles, or 49.5 g) to a farm field and sampling their breakthrough curves at the outlet to a stream, approximately 30 meters away. Simultaneously, we performed a 26-hour time-lapse electrical resistivity tomography (ERT) survey to monitor the saline tracer migration in three dimensions every 0.5 to 1 hour. The initial pulse of tracers had a mean arrival time of 21 minutes and transported 262 g of added Cl (28%), 0.65 g of dissolved P (17%), and 1.4x10¹⁰ particles (4%) to the tile drain outlet. A stochastic mobile-immobile model fit the anomalous (non-Fickian) solute breakthrough curves, where the mobile zone represents the macropore and tile drain network, and the immobile zone represents the soil matrix. Residence times in the immobile zone exhibited a heavy (power-law) tail. ERT images confirmed the retention of tracer mixture in soils after concentrations were no longer measurable at the tile drain outlet. Core samples suggest that 96% of particles and 21% of dissolved P were retained within 10.5 cm of the application location. Solutes and particles were remobilized over longer timescales during three successive storms. Exported masses of Cl and dissolved P at the tile drain outlet ranged from 1,490-12,300 g and 25.7-65.2 g, respectively, indicating flushing of older Cl and P stored in soils before the tracer experiment. Less than 0.01% of the added fluorescent particles were flushed during these storm events. This study indicates the wide range of P travel times through the subsurface in tile drained landscapes and the need to incorporate non-Fickian transport behavior in models.