eLearning

pled rather than absolute changes. This study in-
cluded careful attention to quality assurance, in-
cluding reanalysis of archived soils when analytical
methods changed from Kjeldahl-N to furnace com-
bustion methods (furnace-combustion analysis av-
eraged approximately 8% more N recovery than
Kjeldahl-N analysis).
Four permanent plots were sampled at each point
in time in watershed #2. The authors concluded
that soil N differed substantially among years, based
on an analysis of variance that apparently treated
each year as an independent, class variable. They
concluded that although N changed significantly
over time, “. . . there was not a strong declining
trend during the 20-year period.” They apparently
did not perform statistical test for any trend over
time (with time as a continuous variable); such an
analysis with their data (assuming a constant bulk
density of 1.0 kg/L) showed a highly significant (P
, 0.01) decline of 60 kg N ha-1
y-1
in the 0–30-cm
mineral soil (Figure 2). The 8% greater N recovery
from furnace-combustion analysis relative to Kjel-
dahl digestions would add approximately another
10 kg N ha-1
y-1
to this net rate of N loss (that is,
Kjeldahl-N in 1994 would be approximately 200 kg
N/ha higher than the furnace N). Where did the
missing 60 to 70 kg N ha-1
y-1
go? Knoepp and
Swank (1997) estimated that approximately 13 kg
Nha-1
y-1
of this missing N may have accumulated
in vegetation and the O-horizon.
The authors stated that a discussion of errors of
estimation was beyond the scope of their study but
suggested that perhaps N was accumulating over
time in the mineral soil below 30 cm depth. This
suggestion is not supported by other research on N
fluxes at Coweeta. For example, the sum of ammo-
nium, nitrate, and organic-N leaching from the BA
horizon into the BC-horizon in this same watershed
was estimated to be only 0.6 kg N ha-1
y-1
(Johnson
and Lindberg 1992), or two orders of magnitude
less than the apparent disappearance from the 0–30
cm mineral soil. The output of ammonium plus
nitrate in streamwater averaged only 0.04 kg N ha-1
y-1
for this watershed, much less than the input of
approximately 5 kg N ha-1
y-1
(Swank and Waide
1988). Denitrification losses should have been less
than1kgNha-1
y-1
(Davidson 1986). These fluxes
sum to a removal of approximately 13 kg N ha-1
y-1
into vegetation and O-horizon, and approximately
1kgNha-1
y-1
in leaching losses, plus approxi-
mately 5 kg N ha-1
y-1
added back to the system
from atmospheric deposition, for an overall expec-
tation of approximately 9 kg N ha-1
y-1
removed
from the mineral soil. This leaves approximately
50–60kgNha-1
y-1
of unaccounted losses. Either
the apparent loss of N was not real or the vector of
loss is not consistent with other N fluxes estimated
for the intensively studied watersheds at Coweeta.
The Hubbard Brook Sandbox Forests
F. H. Bormann and others (1977) summarized ni-
trogen pools and fluxes in 55-year-old hardwood
forest of watershed 6 at the Hubbard Brook Exper-
imental Forest in New Hampshire, USA. The vege-
tation was dominated by sugar maple (Acer saccha-
rum L.), American beech (Fagus grandifolia Ehrh.),
and yellow birch (Betula alleghaniensis Britton).
They estimated atmospheric deposition, N incre-
ment in trees, and streamwater output. They as-
sumed the mineral soil could not be a net source of
N and then concluded that an occult source must
supply approximately 14 kg N ha-1
y-1
.
Although occult N inputs may seem less likely
than a net change in the N content of the mineral
soil, the assertion of substantial input of occult N
spurred the later “sandbox” experiment to test for
occult N inputs under very controlled conditions
(BT Bormann and others 1993). The sandboxes
were 7.5 x 7.5 m, lined with Hypalon polymer,
filled to a depth of 1.5 m with 15 cm gravel (at the
bottom for aeration), 130 cm of glacial outwash
sand (low in N), and 5 cm of topsoil. One sandbox
was planted at high density with pitch pine, and
another with red pine. After 5 years of tree growth
(soil sampling started after 1 year), the authors
estimated the N and C content of the soils and the
vegetation. An unvegetated box lost both N and C
(no change in C:N), the locust and alder sandboxes
gained N but not C, and the pine sandboxes lost N
but not C (Figure 3). The varied sampling schemes