Development of Vegetation Models to Predict the Potential Effect of Groundwater Withdrawals on Forested Wetlands

We developed vegetation models that, when linked to groundwater-hydrology models and landscape-level applications, can be used to predict the potential effect of groundwater-level declines on the distribution of wetland-forest communities, individual wetland species, and wetland-indicator groups. An upland-to-wetland vegetation gradient, comprising 201 forest plots located in five different study basins and classified as either upland pine-oak, pitch pine lowland, pine-hardwood lowland, hardwood swamp, or cedar swamp, paralleled variations in water-level. Water levels, woody-species composition, the percentage of wetland- and upland-indicator species, and soil properties varied among the five vegetation types. Because of the functional relationship of hydrology with its correlated soil variables, hydrology represented a good proxy for the complex hydrologic-edaphic gradient associated with the upland-to-wetland vegetation gradient. Two types of vegetation models were developed to predict potential changes in vegetation associated with water-level declines. Logistic regression models predicted the probability of encountering the different vegetation types and 29 community-indicator species in relation to water level. Simple regression models predicted the relative abundance and richness of wetland-and upland-indicator species as a function of water level.