Riparian wetlands are under heavy pressure from hydrological changes produced by dam construction and water diversion projects. There has been ample documentation of the relationship between the extent of flooding and the composition of shoreline plant communities, yet we have few models that allow us to predict the impact of altered flooding regimes on riparian wetlands. In the humid temperate zone, river regulation commonly affects the distribution of two major vegetation types: wooded wetland and herbaceous wetland. The practice of reducing peak floods and augmenting minimum river flows is often followed by the succession of herbaceous to wooded wetland. We used logistic regression models to describe the distribution of wooded wetland as a function of all possible combinations of seven hydrological variables. The variables were chosen to reflect the depth, duration, and time of flooding and were calculated for four different time intervals (3, 7, 12, and 18 growing seasons). Our best model was a combination of two variables: the last day of the first flood and the time of the second flood. For three of the four time intervals, the vegetation type was correctly identified as herbaceous or wooded for Ͼ80% of the sample points. Our results suggest that models based on a few key environmental variables can be valuable tools in the conservation management of the vegetation of temperate and boreal zone wetlands.

Reference wetlands play an important role in efforts to protect wetlands and assess wetland condition. Because wetland vegetation integrates the influence of many ecological factors, a useful reference system would identify natural vegetation types and include models relating vegetation to important regional geomorphic, hydrologic, and geochemical properties. Across the U.S. Atlantic Coastal Plain, depression wetlands are a major hydrogeomorphic class with diverse characteristics. For 57 functional depression wetlands in the Upper Coastal Plain of South Carolina, we characterized the principal vegetation types and used a landscape framework to assess how local (wetland-level) factors and regional landscape settings potentially influence vegetation composition and dynamics. Wetland sites were stratified across three Upper Coastal Plain landscape settings that differ in soils, surface geology, topography, and land use. We sampled plant composition, measured relevant local variables, and analyzed historical transitions in vegetative cover types. Cluster analysis identified six vegetation types, ranging from open-water ponds and emergent marshes to closed forests. Significant vegetation-environment relationships suggested environmental "templates" for plant community development. Of all local factors examined, wetland hydrologic regime was most strongly correlated with vegetation type, but depression size, soil textural type, and disturbance history were also significant. Because hydrogeologic settings influence wetland features, local factors important to vegetation were partly predictable from landscape setting, and thus wetland types were distributed non-randomly across landscape settings. Analysis of long-term vegetation change indicated relative stability in some wetlands and succession in others. We developed a landscape-contingent model for vegetation dynamics, with hydroperiod and fire as major driving variables. The wetland classification, environmental templates, and dynamics model provide a reference framework to guide conservation priorities and suggest possible outcomes of restoration or management.

Riparian wetlands are under heavy pressure from hydrological changes produced by dam construction and water diversion projects. There has been ample documentation of the relationship between the extent of flooding and the composition of shoreline plant communities, yet we have few models that allow us to predict the impact of altered flooding regimes on riparian wetlands. In the humid temperate zone, river regulation commonly affects the distribution of two major vegetation types: wooded wetland and herbaceous wetland. The practice of reducing peak floods and augmenting minimum river flows is often followed by the succession of herbaceous to wooded wetland. We used logistic regression models to describe the distribution of wooded wetland as a function of all possible combinations of seven hydrological variables. The variables were chosen to reflect the depth, duration, and time of flooding and were calculated for four different time intervals (3, 7, 12, and 18 growing seasons). Our best model was a combination of two variables: the last day of the first flood and the time of the second flood. For three of the four time intervals, the vegetation type was correctly identified as herbaceous or wooded for >80% of the sample points. Our results suggest that models based on a few key environmental variables can be valuable tools in the conservation management of the vegetation of temperate and boreal zone wetlands.

No abstract