LAUSR

Lake-edge wetland types result from the influence of various hydrogeomorphological (HGM) predictors, such as water level, hydroperiod, and elevation. Climate and land-use changes could induce variations in lake water levels, and subsequently trigger changes in wetland type and area. This study aims to determine whether HGM predictors alone can be used to classify lake-edge wetlands. The resilience of these wetlands to different water level change scenarios was also evaluated. We sampled the plant communities of 37 lake-edge wetlands (peatland, ash and alder swamps) of a medium-size Canadian lake (12.9 km2), and computed 12 HGM predictors using a GIS software and commonly available landscape data. We used canonical redundancy analysis to evaluate the relationship between wetland types, plant communities, and HGM predictors. We then built a random forest model using these predictors to classify wetlands and to evaluate the impacts of small (± 0.5 m) and large (± 2 m) water level changes on wetland type and area. Our results show that elevation and slope were the two main HGM predictors of wetland plant communities and types. The random forest model was robust (89 % accuracy), but performed less well for swamps, likely because alder and ash swamps are similar systems, but at different successional stages. Finally, while small water level changes had minimal impacts on lake-edge wetland type and area (−0.6 %), larger changes triggered substantial gains (+ 13 %) or losses (−20 %) of tree-dominated wetlands. Considering that swamps harbour high species richness, landscape management should be carried out bearing in mind their greater vulnerability.