The hypothesis that collembola affect rhizobia and mycorrhizas of soyb
ean (Glycine max) and thus indirectly change leaf tissue nutrient conc
entration was studied in pot and field experiments. When a high densit
y of the collembolan species Folsomia candida, was added to pots, the
number of nodules per plant increased 52%. When moderate densities of
two collembolan species, Folsomia candida and Tullbergia granulata, we
re added in factorial combinations to cylinders sunk in the soil aroun
d soybean in fields, the following responses were observed: 40% greate
r mycorrhizal root length, and 5% higher leaf tissue N, but no changes
in leaf P, nodule number or root mass. Collembola density in the fiel
d was too low to increase nodule number per plant as observed in pot e
xperiments: there was no mechanism to explain the 5% increase in leaf
tissue N associated with collembola in the field. In the field, interm
ediate densities of collembola were associated with greater mycorrhiza
l root length, bur since available soil P concentrations were high, lo
nger mycorrhizal root length was not associated with higher leaf tissu
e P. A path model showed that if mycorrhizas had been positively assoc
iated with higher leaf P, the indirect effect of collembola would have
been significantly higher leaf tissue P. This study showed that both
available soil P and collembola density determine mycorrhizal benefits
. In natural habitats, intermediate collembola density and low soil P
are expected to maximize benefits of mycorrhizas to plants.