Mixed model ANCOVA for M. polymorpha grown either in the presence or absence of E. medicae in non-serpentine or serpentine destination soils in the greenhouse1.

1Only the fixed effects or interactions involving presence/absence of rhizobium are shown; random effects reflecting the main plot, split-plot and split-split-plot levels of organization of the factorial experimental design were not of experimental interest. Bonferroni correction for two hypotheses per test yields alpha = 0.025 as a threshold for significance; F-statistics in bold are significant. †, Pb, Pc, Pd, P