Roots are widely acknowledged to be major contributors to ecosystem nu
trient cycles. However, live roots may have very different effects tha
n dead roots, the quality of the soil surrounding the roots may itself
affect the influence of the roots, and the nature of the soil profile
may modify all of these interactions. We have used experimental, fiel
d-based mesocosms to dissect the differing effects of live and dead ro
ots in organic and mineral horizons of spodosolic forest soils, Mesoco
sms were constructed in a simple ecosystem dominated by pitch pine (Pi
nus rigida Mill.) and blueberry (Vaccinium corymbosum L.) in the New J
ersey Pinelands. They contained 5 cm deep surface horizons of either o
rganic matter or an inert substitute (vermiculite), placed above 10 cm
of mineral material from the E horizon, with defined amounts of live
or dead roots added. Sampling was conducted over a 2-yr period. In min
eral soil, live roots stimulated ammonification rates but had little e
ffect on extractable N, and the stimulation occurred only if an organi
c surface horizon was present. In contrast, live roots in organic mate
rial reduced both ammonification rates and extractable N. Dead roots i
ncreased extractable N in the mineral material, but not the organic ma
terial. The presence or absence of an organic horizon had a dominating
effect on all aspects of nitrogen dynamics in the mineral soil; not o
nly extractable inorganic N, but also transformation rates, were highe
r in soils under organic matter. The results showed that the effects o
f roots on soil nutrients depended on (1) the ratio of live to dead ro
ot biomass, (2) the quality of the soil material in which the roots oc
curred, and (3) the structure of the soil profile. These findings nece
ssitate a revision of the concept of ''rhizosphere effect'' in forest
soils.