Results of the statistical models.

The dependent variables were seedling survival, modeled as the survival success out of 5 trials with a binomial distribution in Generalized Linear Model, and seedling growth rate (mg day-1) modeled with a normal distribution in a General Linear Model. We could not analyze growth rate of C. rotundifolia, H. radicata, N. stricta and R. acetosa as there were too few surviving seedlings in at least one treatment. Nitrogen addition, phosphorus addition and clipping treatment and their interaction factors were the explanatory variables.