Modeling wetland vegetation community response to water-level change at Long Point, Ontario

Three spatially-explicit wetland models were developed in a geographic information system (GIS) to simulate wetland vegetation response to water-level fluctuations at the Long Point, Ontario wetland complex. They included: a rule-based model that used a series of if-then statements related to pre-existing vegetation, water depth and wetland vegetation community tolerance ranges; a vegetation state probability model based on likelihood of certain wetland vegetation communities occurring at specific water depths; and a vegetation transition probability model based on likelihood of wetland communities changing to another community under declining or rising water level conditions. The accuracy of the models was evaluated by comparing area and spatial distribution of the simulated wetland landscape to digital historical wetland vegetation data from air photo interpretation. The accuracy of the models ranged from over 80% of the cells correctly classified by the vegetation transition probability model and rule-based model to about 55% correctly classified by the vegetation state probability model. The vegetation transition probability model was marginally more accurate than the rule-based model when assessed on a cell-by-cell basis, but the rule-based model replicated the spatial distribution of vegetation communities more accurately and may be more broadly applicable. Recommended improvements include: additional environmental factors (wave exposure and substrate) incorporated in the decision rules and more detailed input data for the digital elevation model (DEM). Spatially-explicit modeling such as the rule-based model can explore management issues related to climate change and water-level regulation impacts on wetlands in the Great Lakes basin and elsewhere. Crown Copyright (C) 2013 Published by Elsevier B.V. on behalf of International Association for Great Lakes Research. All rights reserved.