Decision rules for riparian vegetation: Predicting the distribution of plant functional groups from trait-environment relationships

The conservation of biological diversity requires simple, predicative models that yield quantitative guidelines for management. This is illustrated by contemporary problems in managing wetlands. Riparian wetlands, in particular, are under heavy pressure from the hydrological changes produced by dam construction and water diversion projects. There has been ample documentation of the significance of substrate and hydrologic variables on the composition of shoreline plant communities, yet we have few models that yield quantitative predictions for conservation management. My goal has been to provide quantitative decision rules that predict the composition of wetland plant communities.; A principal obstacle to generating predictions for wetland vegetation is the large species pool. One solution is the assignment of species to functional groups that are recognizable by a few, simple traits. The distribution of these functional groups could then be described by trait-environment relationships. I have investigated the relationship between the distribution of functional groups and a set of hydrological and substrate variables, using the wetlands of the Ottawa River as my study system. I first identified two fundamental functional groups of wetland species, woody and herbaceous. The model with the highest accuracy was composed to two hydrologic variables, the last day of the first flood and the time of the second flood. The combination of these two variables correctly identified which group was the cover type for over 80% of the original data points.; I then investigated the relationships between functional groups defined by the presence or absence of floating leaves and/or flexuous stems and the environmental variables. The duration of flooding best described the distribution of this group, again with an accuracy of over 80%, using the original data points.; Finally I compared the merits of five classifications of emergent species in identifying functional groups that can be predicted from hydrological and substrate variables. The traits used in these classifications included leaf shape, area or arrangement; stem characteristics; ramet biomass; ramet height; and the number and length of rhizomes. The most accurate models (based on the original data) related the distribution of the functional groups that were defined by leaf shape, leaf area and stem diameter to hydrology and the sand fraction in the soil.; I generated and tested quantitative predictions from the 11 models with the highest accuracy (using the original data) from the above work. Four models performed well when tested with data from other regions or from Ottawa River sites not previously included in the analysis. The corresponding functional groups were woody plants, floating leaved/flexuous stemmed plants, plants with leaf area equal to zero and plants with lanceolate, elliptical or compound emergent leaves. Using the model parameters I formulated inclusion and exclusion rules to predict the presence or absence of these functional groups at various ranges of the significant environmental variables. These decision rules provide valuable tools for the conservation management of the vegetation of riverine wetlands.