Figure 1.

Experimental design. Phylogenetic classification was used to establish a balanced design with litter mixtures created as follows: single-species, within-genus, within-family and between-family. Note that a set of core species is present at all levels of phylogenetic distance each of which is mixed with successively more distantly related species. This design minimizes resampling of phylogenetic branches in different mixtures as it would occur if all combinations of all species were included. The analysis then accounts for the identity of species included in a litter mixture by calculating overyielding, e.g. mass loss of the species mixtures relative to mass losses of each of the species in isolation. Note also that each replicate of a given phylogenetic distance is a distinct species combination, avoiding pseudoreplication within combinations. Age estimates were based on the genus age-optimized tree and an alternative tree was analysed in addition (Material and methods).