A typical least squares mixed-model ANOVA table of net annual aboveground biomass production in grassland biodiversity experiments.

Species richness (continuous, log2 scale) is the only fixed effect. The R2 - the percentage of the total sum of squares explained by each row of the table - is a limited measure of relative importance because it does not account for the large differences in degrees of freedom (DF). F or P values do not indicate relative importance because different explanatory variables are tested against different error terms (column 7): (1) Experiment (random) is tested against block; (2) Block (random) against the overall (between plot) residual error; (3) Species richness (fixed) against species composition; (4) Experiment*Richness (random) against Experiment*Composition; (5) Species composition (random) against its interaction with experiment; (6) Experiment*Composition (random) against the overall (between plot) residual error (following Hector et al. 1999 and Spehn et al. 2005).