Dynamics of P-binding forms in sediments of a mesotrophic hard-water lake: Insights from non-steady state reactive-transport modeling, sensitivity and identifiability analysis

Sediments can act both as a source or sink of contaminants and nutrients in lakes. In this study, we developed a non-steady state reactive transport diagenetic model to gain insights into the dynamics of phosphorus binding forms in the sediments of Lake Simcoe, a mesotrophic hard-water lake located in Southern Ontario, Canada. We investigate three basins of the lake with differences in their phosphorus binding forms, reflecting the distinct spatiotemporal patterns of land use and urbanization levels in the watershed. In the model, total phosphorus is divided into loosely adsorbed phosphorus, phosphorus bound with aluminum, organic phosphorus, redox sensitive and apatite phosphorus, and dissolved phosphorus in pore water. Using the fluxes of organic and inorganic matter along with the concentrations of dissolved substances as dynamic boundary conditions, we simulated the depth profiles of sixteen solute and solid components. The model closely reproduced the fractionation data of phosphorus binding forms. The impact of the interplay between sedimentation fluxes and geochemical conditions on phosphorus diagenesis was then studied under a range of anthropogenic disturbances and natural variability in Lake Simcoe. We also conducted sensitivity analysis that pinpointed the most influential processes underlying the mechanistic foundation of the model. The non-steady state diagenesis model reveals that apatite P dominates the P forms in Cook's Bay, which has been overwhelmingly influenced by agricultural activities in the corresponding watershed during the last 100 years. In contrast, Kempenfelt Bay has been primarily impacted by urbanization and experienced oxygen depletion in the deep water. Thus, we found that organic P binding forms dominated over redox sensitive P when urban loading was intensified. Finally, our study offers insights into the identifiability of the model as well as into the factors that will be critical to monitor in order to improve its credibility. The model outputs are sensitive to the concentrations of dissolved oxygen and pH at the sediment-water interface. The sensitivity with respect to these factors overwhelmingly dominates over all other parameters. Furthermore, the characterization of the sedimentation fluxes; namely, the composition of settling organic matter, expressed as the ratio of degradable to inert organic matter is the second strongest factor that can influence the inference drawn by our modeling exercise. Crown Copyright (C) 2013 Published by Elsevier B.V. All rights reserved.