eLearning

soil N (and C) at a watershed scale, in a rocky soil,
after clear-cutting. With 60 carefully sampled soil
pits in 1983, Johnson (1995) found a total N con-
tent of the soil (O-horizon to C-horizon) of 7200 kg
N/ha, with a standard error of the estimate of 340
kg N/ha. Three years after harvesting, the soil N
appeared to increase to 7600 kg N/ha, but a change
of 400 kg N/ha was not close to significant. The
third sampling (in 1991) yielded an N content of
6700 kg N/ha, but this was not significantly differ-
ent from the original estimate (P , 0.3). Johnson
concluded that no change in soil N could be de-
tected within any soil horizon or the whole soil
profile. However, the large variance (which was
unavoidable in the rocky soil) would prevent de-
tection of any rate of change that was less than a
couple hundred kg N ha-1
y-1
. Continued sampling
over longer periods will lower the possible rate of
annual change that could escape detection.
Postharvest Change in Soil N in Upper
Michigan
Mroz and others (1985) sampled soils at three lo-
cations before harvesting, and 1.5 years later. At
each sampling period, five soil cores were taken in
each of 24 plots in each of the three stands, a
sampling intensity rarely used in studies of forest
soils. The average change in N content over 2.5
years was 1320 kg N/ha (95% confidence interval
633 kg N/ha based on three site averages). Virtu-
ally all of this loss occurred in the first 1.5 years,
with no further loss in the following year (Mroz and
others 1985). The loss of N included up to 80% of
the forest floor and much of the N in the mineral
soil. This rate of change in N is the highest ever
reported for a forest soil and if true would have
major effects on groundwater N (yielding average
concentrations 10 times greater than drinking wa-
ter standards), the H1 budget, and long-term soil
fertility (Binkley 1987). An undetermined portion
of the N may have been moved around on the site
during the harvesting operation rather than lost in
leachate or as a gas (Mroz and others 1987). This
study has not been repeated to verify the magni-
tude of the loss, or to determine the process that
could account for such huge fluxes. Studies in this
region that directly assessed leaching losses of N
after harvest have found maximum increases of less
than 10 kg N ha-1
y-1
(see Richardson and Lund
1976; Iseman and others 1999).
Prescribed Fire in a Southern Pine Forest
Jorgensen and Wells (1971) reported a rate of N
accretion of 23 kg N ha-1
y-1
over a period of 10
years in 0–10 cm of mineral soil in replicated plots
of loblolly pine that were burned annually over 10
years in the lower Coastal Plain of South Carolina.
No variance among plots was given. However, more
complete data presented by Wells (1971) showed
that the fire increased the N content of 0–10-cm
mineral soil in direct proportion to the decrease in
N in the O-horizon, showing no net change across
all horizons. Binkley and others (1992) reported a
similar pattern, except that higher N in the mineral
soil could not account for all the reduced N content
of the O-horizon.
SUMMARY OF OCCULT NRATES
The frequencies of reported rates of N accretion in
forests are graphed in Figure 4, along with our
evaluation of the strength of the experimental de-
sign behind each estimate. As noted above, the
estimates come from a variety of situations, meth-
ods, and extent of sampling. The graph omits the
three studies that reported significant net losses of N
(Hubbard Brook unvegetated sandbox; Coweeta
watershed #2; and Hardwood Harvesting study in
Michigan). The Walker Branch Watershed study
that found an average N loss of 35 kg N ha-1
y-1
is
listed in Figure 4 in the 0–5 kg N ha-1
y-1
class
because the rate was not significantly less than 0. Of
the 25 reports we found of N accretion in forests,
approximately half reported rates of 10 kg N ha-1
y-1
or less, giving no evidence of occult N inputs. Of