Statistical classification of vegetation and water depths in montane wetlands

Relationships between water depths and density of submergent vegetation were studied in montane wetlands using statistical techniques based on clustering and an extension of regression trees. Sago pondweed (Stuckenia pectinata) was associated with lower average water depths than water milfoil (Myriophyllum sibiricum). We detected a nonlinear relationship when average water depths were used to predict percent cover in S. pectinata, with depths of 30–40 cm, producing the highest predicted average percent cover of S. pectinata; higher and lower depths resulted in lower percent cover predictions. For M. sibiricum, higher water depths were monotonically associated with higher average percent cover. To foster more S. pectinata and less M. sibiricum, managers might employ water control structures to reduce water depths below 1 m, using both temporary drawdowns and average depths of 30–40 cm. Other species responded less markedly to water depth variation. Should decreased water depths become more common, these results suggest an increase in S. pectinata and a decrease in M. sibiricum.