Nutrient-driven plant invasions in wetlands around the Michigan coastline: Using satellite and field data to test a model linkage across scales

Background/Question/Methods

In the Great Lakes region of the Upper Midwest, USA, agricultural and urban land uses and high N deposition are contributing to elevated flows of N in rivers and groundwater to coastal wetlands.  Coastal wetlands provide a vital, region-wide link between land and water in the Great Lakes Basin.  Elevated N inflows are believed to facilitate the spread of large-stature invasive plants (cattails and Phragmites) that reduce biodiversity and have complex effects on other ecosystem services including wetland N retention and C accretion.  We tested whether a linkage of models across scales from the hydrogeochemistry of large watersheds to local community-ecosystem processes in wetlands could accurately simulate locations of greater plant invasions around the Michigan coastline.  We linked the LHM (Landscape Hydrology Model), applied over the entire Michigan lower peninsula, to the Mondrian wetland model at 8 wetland points around the Michigan coastline.  We used Monte Carlo techniques to estimate invasion risk as a likelihood from 0 to 100% in these wetlands.  We tested these model-linkage results against field observations and statewide-derived remote-sensing classifications of vegetation cover that included the ability to distinguish between native and invasive wetland plant communities.    

Results/Conclusions

At the regional scale, our linked models predicted a general pattern of greater invasion risk in the southern coastlines of lakes Michigan and Huron relative to northern areas where coastal nutrient loading is lower.  At the scale of local wetlands, the linked models predicted threshold behavior in the success of invasive plants in response to N loading, with the threshold ranging from ca. 8 to 12 g N/m2 y, depending on invader species and hydroperiod.  Phragmites invasions had a significant interaction with hydroperiod, in which the nutrient-loading threshold for invasion was significantly lower under aerobic relative to anaerobic (flooded) conditions.  This suggests that if Great Lakes water levels continue to drop as they have since the 1980s, Phragmites invasion risk will increase in coastal wetlands for a given level of nutrient loading, where nutrient inflows are elevated over historical levels.  Satellite remote sensing and vegetation classification produced maps of invasive plants at wetland sites ranging from 0% cover (Thompson’s Harbor) to 92% (Whiskey Harbor).  Modeled plant invasions increased wetland productivity 3-fold over historically oligotrophic native communities and decreased plant biodiversity but slightly increased wetland N retention.