Comprehensive Review of Modeling of Wetland Water Quality and Ecological Processes

Abstract

Wetland ecosystem functions and services are vital to a sustainable environment. Process-based simulation models are powerful tools for testing hypotheses, forecasting wetland ecosystem responses, and designing engineered wetlands for pollutant mitigation. This state-of-the-art review synthesizes water quality and ecological processes with emphasis on their mathematical representation in wetland models. Over the last three decades, process representation has advanced significantly for constructed wetlands, although to a lesser extent for natural freshwater wetlands and coastal salt marshes. The capability of watershed models to simulate wetland processes remains limited; thus, aggregating plot-scale impacts to the basin scale is an imperative area for future research. Biogeochemical and uptake processes are typically simplified into first-order reaction rates within wetland design formulas, often overlooking nonlinear interactions among constituents and rate-limited removal processes. Domain integration can be achieved by combining multiple constituents in Monod-type or empirical formulas—for example, by limiting organic matter mineralization based on nutrient-to-carbon ratios or linking the fates of nutrients, carbon, and metals to oxygen, redox dynamics, and microbial growth. Further research is needed regarding the explicit representation of biogeochemical and biomass growth in variably saturated substrates, microbial biofilm dynamics, and gaseous diffusion and translocation through plant stems (e.g.,  O 2 , CH 4 , N ₂ O, and pesticides). Additionally, limited progress has been reported in modeling the fate and plant uptake of metals, hydrocarbons, pesticides, and emerging contaminants like per- and polyfluoroalkyl substances (PFAS). For coastal salt marshes, models for sequestration and emissions should account for sulfate as an inhibiting oxidizing agent and the lateral exchange of organic carbon to reduce uncertainty. Balancing model parsimony and complexity remains a guiding principle, especially given the resource constraints. Finally, exploring advanced machine learning and probabilistic methods for effective calibration and uncertainty estimation remains a challenging research priority.