Modeling of hydroecological feedbacks predicts distinct classes of landscape pattern, process, and restoration potential in shallow aquatic ecosystems

Abstract

It is widely recognized that interactions between vegetation and flow cause the emergence of channel patterns that are distinct from the standard Schumm classification of river channels. Although landscape pattern is known to be linked to ecosystem services such as habitat provision, pollutant removal, and sustaining biodiversity, the mechanisms responsible for the development and stability of different landscape patterns in shallow, vegetated flows have remained poorly understood. Fortunately, recent advances have made possible large-scale models of flow through vegetated environments that can be run over a range of environmental variables and over timescales of millennia. We describe a new, quasi-3D cellular automata model that couples simulations of shallow-water flow, bed shear stresses, sediment transport, and vegetation dynamics in an efficient manner. That efficiency allowed us to apply the model widely in order to determine how different hydroecological feedbacks control landscape pattern and process in various types of wetlands and floodplains. Distinct classes of landscape pattern were uniquely associated with specific types of allogenic and autogenic drivers in wetland flows. Regular, anisotropically patterned wetlands were dominated by allogenic processes (i.e., processes driven by periodic high water levels and flow velocities that redistribute sediment), relative to autogenic processes (e.g., vegetation production, peat accretion, and gravitational erosion). These anistropically patterned wetlands are therefore particularly prone to hydrologic disturbance. Other classes of wetlands that emerged from simulated interactions included maze-patterned, amorphous, and topographically noisy marshes, open marsh with islands, banded string-pool sequences perpendicular to flow, parallel deep and narrow channels flanked by marsh, and ridge-and-slough patterned marsh oriented parallel to flow. Because vegetation both affects and responds to the balance between the transport capacity of the flow and sediment supply, these vegetated systems exhibit a feedback that is not dominant in most rivers. Consequently, unlike in most rivers, it is not possible to predict the “channel pattern” of a vegetated landscape based only on discharge characteristics and sediment supply; the antecedent vegetation pattern and vegetation dynamics must also be known. In general, the stability of different wetland pattern types is most strongly related to factors controlling the erosion and deposition of sediment at vegetation patch edges, the magnitude of sediment redistribution by flow, patch elevation relative to water level, and the variability of erosion rates in vegetation patches with low flow-resistance. As we exemplify in our case-study of the Everglades ridge and slough landscape, feedback between flow and vegetation also causes hysteresis in landscape evolution trajectories that will affect the potential for landscape restoration. Namely, even if the hydrologic conditions that historically produced higher flows are restored, degraded portions of the ridge and slough landscape are unlikely to revert to their former patterning. As wetlands and floodplains worldwide become increasingly threatened by climate change and urbanization, the greater mechanistic understanding of landscape pattern and process that our analysis provides will improve our ability to forecast and manage the behavior of these ecosystems.