Environ. Sci. Technol. 2005, 39, 7777-7783
Response of the Nitrogen Isotopic
Composition of Tree-Rings Following
Tree-Clearing and Land-Use Change
ANDREW R. BUKATA* AND
T. KURTIS KYSER
Queen’s Facility for Isotope Research, Department of
Geological Sciences and Geological Engineering, Queen’s
University, Kingston, ON, Canada, K7N 3N6
Clear-cutting of forests affects the nitrogen cycle and the
nitrogen isotopic composition of bioavailable ammonium
and nitrate in the soil. Here, we have used nitrogen isotopic
variations of tree-rings in red oak (Quercus rubra) and
white oak (Quercus alba) as indicators of changes in the
nitrogen cycle on a local scale. The δ15N values of latewood from trees at two remnant forest stands in Ontario,
Canada, that underwent large-scale tree-clearing and
permanent land-use change at different times were measured.
Trees from the perimeter of each stand record a marked 1.52.5‰ increase in the δ15N values of their tree-rings
relative to the values in trees from the center of the stand,
with the shift synchronous with the tree-clearing and landuse change. This shift was most likely due to increased
rates of nitrification and nitrate leaching in the soil as a result
of tree-clearing combined with permanent changes in
hydrology and probable fertilizer use accompanying the
change in land-use. Nitrogen concentration in tree-rings
was not affected by tree-clearing and the associated change
in land-use. These results indicate that changes in
nitrogen cycling in forest ecosystems, whether due to
climate change, land-use change, or other environmental
changes (increased O3, other atmospheric pollutants,
insects, etc.), can be faithfully monitored with nitrogen
isotopic compositions of tree-rings and that dendrogeochemical analysis can be incorporated into studies of the
effects of long-term anthropogenic effects on forest
ecosystems.
Introduction
The nitrogen cycle in forested ecosystems has been increasingly disturbed by human activities since the Industrial
Revolution. Manufacturing processes, automobile exhaust,
increased use of fertilizers, changing land-use, and forest
clearing have all disrupted the nitrogen cycle. In forested
ecosystems where nitrogen is most often a limiting nutrient,
increased atmospheric nitrate and ammonium deposition
and perturbations to the nitrogen cycle can have adverse
effects on forest health (1, 2). Addition of nitrogen to a
normally nitrogen-limited system disturbs biogeochemical
cycles and causes the system to seek a new steady state.
Annual growth rings combined with reasonable longevity
make trees possible long-term sentinels of such nitrogen
cycle disturbance in forested ecosystems.
* Corresponding author phone: (613)533-2183; fax: (613)533-6592;
e-mail: [email protected].
10.1021/es050733p CCC: $30.25
Published on Web 09/07/2005
 2005 American Chemical Society
FIGURE 1. Schematic of the nitrogen cycle in forests modified from
Nadelhoffer and Fry (1994) (21). Arrows indicate the direction of
kinetic nitrogen transformations. The ∆ values (δ15Nproduct - δ15Nsource)
next to the arrows indicate the range of expected changes in the
δ15N value of the product as summarized in the literature (4, 21).
The dashed box inside the plant reflects internal nitrogen cycling
within the plant. Clear-cutting has been shown to increase the
rates of nitrification, denitrification, and nitrate leaching.
The δ15N value of nitrogen that is bioavailable to a tree
will be affected by (1) the introduction of an isotopically
different nitrogen source and (2) altering the processes that
make nitrogen available for uptake (3, 4). The δ15N value of
foliage has been used to examine the nitrogen cycle in
undisturbed forests (5-7), in forests receiving fertilization
(8-11), and to determine the effects of clear-cutting on the
nitrogen cycle of forests (12, 13). Changes in the δ15N value
of source nitrogen have been shown to affect the δ15N value
of foliage in fertilization (8, 9) and atmospheric nitrogen
pollution studies (14) and as a result of salmon carcass
deposition by bears (15). In each study, the introduction of
nitrogen from a source with a δ15N value distinct from the
existing foliage at the site was recorded by a shift in the foliar
δ15N value. Only a few studies (16-19) have attempted to use
the δ15N values of tree-rings to infer temporal changes in the
nitrogen cycle.
Poulson et al. (1995) (16) reported a decrease in the δ15N
values of tree-rings of two eastern hemlocks (Tsuga canadensis) after 1960 that they attributed to anthropogenic
nitrogen emissions. Peñuelas and Estiarte (1997) (17) found
a 0.14‰ decrease in δ15N values of downy oak (Quercus
pubescens) over the last century. Hart and Classen (2003)
(18) sampled ponderosa pine (Pinus ponderosa) that had
been labeled with an 15N-enriched tracer 14-years earlier
and found that the tree-rings recorded the event but that
there had been some radial translocation of the tracer. Saurer
et al. (2004) (19) reported that both the nitrogen concentration
and the δ15N values of tree-rings from Norway spruce (Picea
abies) increased proximal to and coincident with construction
of an adjacent highway. They attributed the increases to
vehicle emissions.
This investigation attempts to determine whether changes
in land-use associated with permanent tree-clearing can
produce changes in the nitrogen cycle that are recorded in
the δ15N values of tree-rings from red oak (Quercus rubra)
and white oak (Quercus alba), two species that preferentially
use nitrate rather than ammonium. Clear-cutting has been
shown to cause increased rates of nitrate leaching, increased
denitrification (1, 2), and increased productivity of nitrifying
bacteria (20). Bacterial nitrification of ammonium to nitrate
is accompanied by an isotopic fractionation that results in
the remaining ammonium having a higher δ15N value than
the initial value (4, 21) (Figure 1). As the microbially available
soil ammonium pool is nitrified, the δ15N value of the nitrate
VOL. 39, NO. 20, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7777
produced will increase because this nitrate is produced from
an ammonium pool with an increasing δ15N value. As the
nitrate lost by leaching is replaced with newly produced
nitrate, the δ15N value of bioavailable nitrate in the soil will
increase. Increased denitrification will also act to increase
the δ15N value of bioavailable nitrate in the soil by preferentially removing 14N to the atmosphere. If the land-use
change includes the application of fertilizers, a new source
of potentially isotopically different nitrogen is introduced to
the system after tree-clearing. By increasing the productivity
of nitrifying bacteria, increasing nitrate leaching, and increasing denitrification, clear-cutting forests should result
in an increase in the δ15N values of ammonium and nitrate
available for uptake by trees. Application of fertilizer to the
clear-cut area will affect the nitrogen cycle of the remnant
forest. Both the amount of fertilizer used and the δ15N value
of the fertilizer may affect the δ15N values of the tree-ring
record if the fertilizer migrates into the forest. Clear-cutting
also removes future litter-fall at the site, which has a lower
δ15N value than most soil nitrogen (21). The long-term
elimination of 15N-depleted litter inputs to the soil effectively
increases the δ15N values of the residual soil organic matter
available for mineralization. The soil nitrogen cycle is not a
closed system, and perturbation caused by tree-clearing and
land-use change can be expected to move the system toward
a new steady-state equilibrium. In this study, we examine
the shift in δ15N values of tree-rings at two geologically distinct
sites where deforestation occurred at different times.
Experimental Section
Study Areas. Trees examined in this investigation were
sampled from sites located in Burlington and Kingston,
Ontario, Canada, on the north shore of Lake Ontario, ∼400
km apart. The Burlington site is a remnant second-growth
forest running along a stream that was left when the
surrounding land was cleared to construct a housing
subdivision in the early 1970s. The soil on the site is a welldeveloped mineral soil (Gray Brown Luvisol) above shale
and carbonate bedrock. Three trees from the center of the
stand were selected, one ∼150-years old, and two ∼70-years
old red oaks (Quercus rubra) spaced 20 m apart. To minimize
the effects of the construction, these samples were collected
at least 10 m from the current perimeter of the forest. In
addition to the trees from the center of the stand, three red
oaks were sampled from the perimeter of the stand, immediately adjacent to a field cleared for the construction of
a high school completed in 1976. All three perimeter trees
were approximately 70-years old.
The Kingston site was a small (∼300 × 500 m) remnant
second-growth forest bound by Lake Ontario to the south,
a two lane road to the north, and bordered by cottage style
homes to the east and west. Three white oaks (Quercus alba)
from the center of the stand were sampled. One was ∼150years old, and the other two were both approximately 120years old. The trees are less than 10 m apart and rooted in
thin (<2 m deep) mineral soil (Gray Wooded Orthic Podzol)
above limestone bedrock. A white oak was sampled from the
east perimeter of the stand where landscaping in 1990
established the perimeter of the forest. The Kingston
perimeter tree was felled by a regional ice storm in February
of 1998, when the tree was ∼43-years old. The same ice storm
also felled an ∼65-years old slippery elm (Ulmus rubra) from
the west perimeter of the stand. The elm became a perimeter
tree in 1970 when land was cleared for construction of cottage
style homes.
Materials and Methods. Sample collection from living
trees was done using a 5 mm-diameter increment corer. The
corer was rinsed with acetone followed by deionized water
(DI-water) between trees, and flushed with DI-water between
each core. For nitrogen concentration and isotopic com7778
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 20, 2005
position, four cores were taken at breast height on each tree,
one each at 90° intervals around the tree. Two additional
cores from opposite sides of each tree were taken for ring
width analysis. These cores were air-dried, mounted, and
sanded prior to having their ring widths measured on a
Velmex stage in the Department of Geography at Queen’s
University. From these measurements, the basal area increments (BAI) were calculated. From the Kingston trees felled
by the ice storm, during clean up, a disk corresponding to
breast height was removed from each tree, from which core
was extracted at 90° intervals. Ring widths were measured
on the disk prior to core extraction.
Cores taken for nitrogen analysis were dissected into
annual growth rings with a stainless steel blade, and the
growth rings were further separated into earlywood and
latewood. Unless otherwise indicated, the annual rings from
each of the four cores from each tree were grouped to create
a single sample for each year. From the latewood sample, a
subsample of 7-25 mg was weighed into a tin capsule and
analyzed by EA-CF-IRMS using either a Carlo Erba NCS 2500
Elemental Analyzer coupled to a Finnigan MAT 252 IRMS or
a Costech ECS 4010 coupled to a Finnigan MAT Delta Plus
XP in the Queen’s Facility for Isotope Research. The nitrogen
concentration and δ15N value of every even numbered year
were analyzed. Precision and accuracy of nitrogen concentrations were determined using multiple analyses of selected
samples, NBS-1547, NBS-1577b, and a white oak sapwood
laboratory standard, and indicate a relative error of 3%. The
δ15N values, which are reported in units of per mil (‰) (22)
relative to air (δ15N ) 0‰), were calibrated using ammonium
sulfate RM-8548 (δ15N ) +20.3 ( 0.2‰), RM-8550 (δ15N )
-30.4 ( 0.5‰), and the laboratory standard and indicate an
uncertainty of 0.3‰ (2σ). Excess O2 was used to ensure all
carbon was converted to CO2 with negligible CO production,
and the CO2 removed using an in-line CO2 scrubber (ascarite
or Carbosorb). Prior to introduction to the mass spectrometer,
the sample gas passed through a GC column that completely
separated CO, N2, and CO2. The low nitrogen concentration
(0.05-0.30 wt %) and high C, O, and H content of the samples
required the CO2 trap to be changed after each run (∼50
samples) and the water trap to be changed after every four
runs. Based on repeated measurements throughout each run,
the blank contribution was assessed to be minor. Acceptable
data were obtained when at least 0.005 mg of nitrogen per
sample was introduced to the source.
To assess the effect of water labile nitrogenous compounds
on the analyses, five 0.5-2.0 g samples from each of three
segments in the heartwood (1955-64, 1970-75, and 198589) and one segment in the sapwood (1991-97) of the ∼43years old perimeter white oak from the Kingston site were
placed into separate vials with 5 mL of DI-water. The samples
were periodically shaken, and the water was removed and
replaced with fresh DI-water four times a day for 3 days,
thereby washing the samples of labile nitrogenous compounds. After 3 days, the washed samples were then airdried in a 40 °C oven and analyzed along with five untreated
samples from each segment of the core.
The difference in nitrogen concentration and δ15N values
of earlywood and latewood from the same growth ring was
assessed using earlywood and latewood fractions of 47 annual
rings from the ∼150-years old red oak from the center of the
Burlington stand. In a ring porous species such as red oak,
the earlywood is formed prior to bud-break and contains
large vessels so the boundary between earlywood and
latewood is sharp and easily identifiable. The nitrogen
concentration and δ15N values of the earlywood and latewood
from the same growth ring were compared.
Radial variability of nitrogen in the bole was assessed
using one ∼120-years old white oak from the center of the
stand at the Kingston site and the ∼65-years old slippery elm
FIGURE 2. (a) Schematic showing the transect cores around the perimeter elm at the Kingston site. (b) δ15N values of tree-rings from
each transect from (a). Tree-clearing occurred in 1970, and three time periods are identified. Also shown is the pre-event average (1σ.
(c) δ15N values of four transects taken from one of the ∼120-years old white oaks from the center of the stand.
(Ulmus rubra) from the west perimeter of the Kingston site.
From the breast height disk of wood removed from the
perimeter elm, four pith to bark samples were taken, one
facing into the clearing, one facing into the remaining stand,
and two parallel to the clearing perimeter (Figure 2a). Five
growth rings from before tree-clearing and seven from the
period after the clearing were selected, and the growth ring
on each transect was analyzed individually. The same years
were sampled from each of the four cores taken from the oak
in the center of the stand. The variabilities in percent nitrogen
and in δ15N values around the tree before and after the treeclearing event were then compared. Selecting a perimeter
tree and a tree from the center of the stand allowed the
variability of each to be compared before and after the treeclearing and land-use change.
Whether the effect of tree-clearing and changing landuse on the nitrogen cycle was recorded in the δ15N values of
tree-rings was examined by comparing trees at the center to
trees at the perimeter of the study stands. At the Burlington
site, three red oaks from the center of the stand and three
red oaks from the perimeter were sampled. The perimeter
trees were all located adjacent to the same field and became
perimeter trees at the same time. The tree-clearing occurred
in 1976, during what is now heartwood in each tree. At the
Kingston site, three white oaks were sampled from the center
of a remnant stand, and one white oak perimeter tree was
sampled. The tree-clearing and land-use change at the
Kingston site occurred in 1990, during the sapwood of the
perimeter tree and the trees from the center of the stand.
This design allowed for the examination of whether a response
in the δ15N values of tree-rings to tree-clearing was seen in
both red and white oak and whether a response in δ15N values
of tree-rings was preserved in both sapwood and heartwood.
Statistical Methods. The extraction experiment results
were compared by paired t-tests following F-tests. The EWLW experiment was compared using a z-test. The radial
variability, both averages and standard deviations, was
compared by ANOVA. The response of δ15N values to landuse change was compared by t-tests comparing the perimeter
tree average to the control tree average for each year for the
Burlington site. For the Kingston site, z-tests were performed.
ANOVA was performed on the average control subtracted
δ15N values from pre-event, post-event, and recovered period
from Burlington and Kingston. Comparing the relative growth
rate of perimeter to center stand trees during the post-event
to a similar time interval immediately prior to the tree-clearing
event was done to assess change in growth rate. The postevent and pre-event ratios were compared by t-test. Statistical
analyses were performed by Microsoft Excel (t-tests, F-tests,
z-tests) and DataDesk (Ithaca, NY) (ANOVA). The level of
significance for all statistical tests was R ) 0.05.
Results and Discussion
Extraction Experiment. Extracting heartwood samples with
DI-water had no consistent significant effect on the weight
percent nitrogen or the δ15N value of the samples (Table 1).
There was no statistically significant difference between the
VOL. 39, NO. 20, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7779
TABLE 1. Results of the Extraction Experiment Including Statistical Analyses
δ15N (‰; air)
heartwood
sapwood
nontreated avg. (1σ)
treated avg. (1σ)
p-value (t-test)
nontreated avg. (1σ)
treated avg. (1σ)
p-value (t-test)
1955-64
1970-75
1985-89
1991-97
3.35 (0.65)
3.88 (0.60)
3.89 (0.44)
4.03 (0.33)
3.86 (0.64)
3.88 (0.78)
4.26 (0.33)
4.62 (0.45)
0.243
0.988
0.166
0.002
0.066 (0.005)
0.065 (0.005)
0.060 (0.004)
0.193 (0.009)
0.069 (0.005)
0.057 (0.005)
0.060 (0.004)
0.167 (0.007)
0.264
0.020
0.718
6.6 × 10-7
δ15N values of treated (by DI-water extraction) and nontreated
heartwood samples. Similarly, the weight percent nitrogen
of treated and nontreated heartwood samples was the same
except in the segment 1970-75 (p ) 0.02). In this case, the
weight percent nitrogen decreased as a result of the extraction, whereas in the other two time-segment samples, the
weight percent nitrogen increased or stayed the same. When
the sapwood was extracted with DI-water, there was a
significant 10.5% decrease in weight percent nitrogen (p <
0.001) and a statistically significant 0.59‰ increase in the
δ15N value (p ) 0.002). Although significant, the δ15N values
are still within the overlap of the analytical errors on both
measurements ((0.3‰).
Nitrogen is a macronutrient, is limiting in most forests,
and is highly mobile within the tree. As a result, it is present
in higher concentrations in living sections of the tree such
as the sapwood (23). The high mobility of nitrogen in the
sapwood is readily demonstrated by the ease of extraction
with DI-water, whereas in the heartwood the remaining
nitrogen compounds in the wood are the least readily
extractable.
Previous workers have argued that chemical extraction is
necessary for accurate measurement of nitrogen in tree-rings,
but these studies were on resinous tree species. Sheppard
and Topa (2002) (24) reported that the isotopic composition
of nitrogen incorporated by the tree in the sapwood of the
resinous species loblolly pine (Pinus taeda L.) reflects the
isotopic composition of the bioavailable nitrogen when all
nonstructural nitrogen-bearing compounds are extracted
prior to analysis. Using the resinous species Ponderosa pine,
Hart and Classen (2003) (18) compared the percent nitrogen
and δ15N values of DI-washed sapwood to holocellulose
extracted sapwood and found a considerably smaller and
less variable effect of the holocellulose extraction than did
Sheppard and Thompson (2000) (25), who did not DI-wash
their samples prior to the extraction. Hart and Classen (2003)
(18) concluded that the holocellulose extraction may not be
necessary when a DI-water extraction is performed. Oak trees,
which we have used in our study, are a nonresinous species.
We have found that in the white oak, DI-water extractions
are not necessary prior to δ15N value measurements.
Earlywood-Latewood Comparison. To remove temporal
trends from the data that may affect average values, differences between earlywood and latewood fractions were
calculated for weight percent nitrogen and δ15N values. We
tested whether the average difference between earlywood
and latewood was significantly different than zero for
heartwood, sapwood, and the entire tree. Earlywood (EW)
contained significantly more nitrogen than latewood (LW)
when all of the growth rings measured were considered (EW
- LW ) 0.047 ( 0.020 wt % N (1σ); p ) 0.018, n ) 47), when
the heartwood alone was considered (EW - LW ) 0.048 (
0.020 wt % N (1σ); p ) 0.015, n ) 42), but not when the
sapwood alone was considered (EW - LW ) 0.039 ( 0.025
wt % N (1σ); p ) 0.113, n ) 5). There was no significant
difference between earlywood and latewood δ15N value when
all of the growth rings were considered (EW - LW ) -0.04
( 1.13‰ (1σ); p ) 0.972, n ) 47), when the heartwood alone
was considered (EW - LW ) 0.00 ( 1.18‰ (1σ); p ) 0.984,
n ) 42), or in the sapwood alone (EW - LW ) -0.33 ( 0.82‰
7780
9
percent nitrogen (wt % N)
sample
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 20, 2005
(1σ); p ) 0.691, n ) 5). The large standard deviations in δ15N
values indicate a high degree of intra-annual variability in
the δ15N value of available nitrogen. Earlywood in oak is
formed during the spring using reserves including nitrogen
stored from the previous growing season (23). As a result, the
earlywood will contain a higher proportion of nitrogen from
internal reserves than will latewood. Rates of snowpack melt,
volume and type of snowmelt runoff, amount of spring
precipitation, as well as spring temperature fluctuations,
which affect the nitrogen cycle in soils, combined with the
use of previously stored nitrogen in earlywood growth
contribute to the intra-annual variation (23). To minimize
the impact of these effects on the long-term record, the δ15N
values of latewood only were used in this investigation.
Nitrogen Concentrations. Nitrogen concentrations in red
oak, white oak, and slippery elm varied from 0.07 to 0.35 wt
%. The percent nitrogen decreased slightly from the pith to
a value between 0.07% and 0.11% throughout the heartwood,
then increased sharply from the heartwood-sapwood transition to a value between 0.20% and 0.35% at the outermost
growth ring. This pattern was present in all 11 trees analyzed.
There were no statistically significant differences between
nitrogen concentrations in average perimeter and average
background tree-rings before and after tree-clearing. The
difference at the Kingston site is near significant (p ) 0.06)
because the trees from the center of the stand were sampled
5 years after the perimeter tree was felled. As a result, the
tree-clearing event was recorded closer to the heartwood in
the trees from the center than the tree from the perimeter
of the stand.
The nitrogen concentrations of heartwood and sapwood
measured in this investigation are similar to those reported
for red oak by Merrill and Cowling (1966) (26). They attributed
the differences in nitrogen concentration to physiological
and biochemical differences between heartwood and sapwood. Trends in nitrogen concentrations in oak in this
investigation are also consistent with those reported in
eastern hemlock (Tsuga canadensis) by Poulson et al. (1995)
(16), which were attributed to tree physiology and were not
related to the amount of nitrogen deposited. Efficient internal
cycling of nitrogen coupled with high mobility suggests that
only the least mobile nitrogen (i.e., structurally bound) would
remain in the wood so that nitrogen concentrations are not
effective tracers of deposition or uptake history.
Sheppard and Thompson (2000) (25) argue for the need
to chemically extract the most labile nitrogen in Ponderosa
pine (Pinus ponderosa) and Douglas fir (Pseudotsuga menziesii) prior to analysis of nitrogen concentrations if the
variations are to be correlated with nitrogen deposition
histories. However, they did not examine the nitrogen isotopic
composition, and therefore did not test the possibility that
the labile nitrogen is isotopically similar to the nonlabile
nitrogen, especially in the heartwood. Although Sheppard
and Topa (2002) (24) found that labile nitrogen compounds
removed from the sapwood are isotopically distinct from
structurally bound nitrogen in a resinous species fertilized
with an isotopically enriched tracer, there is far less nonstructural nitrogen in the heartwood of nonresinous species
such as oak. The DI-water extraction experiments performed
in this investigation found that, although significant amounts
of nitrogen compounds may be extracted, it had no significant
effect on the δ15N value of heartwood or sapwood.
Radial Variability. The Kingston perimeter elm could be
divided into three periods based on the relation between
when clearing occurred and the δ15N values: pre-event
(1934-68) when the δ15N values are low, post-event (197292) when the δ15N values increase significantly, and recovered
(1994-97), when the δ15N values become low again (Figure
2b). The tree-clearing event itself occurred in 1970. The
average δ15N values for each of the three periods are
significantly different (p ) 0.0128). The post-event average
δ15N value (1.81 ( 0.49‰ (1σ)) is significantly different (p )
0.0046) from the pre-event average δ15N value (0.91 ( 0.24‰
(1σ)). The pre-event average and recovered average δ15N value
(1.04 ( 0.21‰ (1σ)) are not significantly different.
One test of radial variability in δ15N values around the
tree is to compare the average standard deviation of each
period from each of the four cores taken. For the perimeter
elm, the average standard deviations for each of the three
time periods are significantly different (p ) 0.0026). The
standard deviation on the average post-event δ15N values
(0.63‰) is significantly greater than the pre-event (0.38‰)
and recovered period (0.42‰). Core taken from the side of
the tree facing the remaining stand (#4 in Figure 2b) had a
smaller increase in δ15N value after the tree-clearing and
land-use change than did core taken from the sides of the
tree facing or parallel to the clearing. Nitrogen concentrations
in the pre-event and post-event time intervals are not
significantly different.
The white oak from the center of the same stand shows
no significant difference in radial variability in δ15N values
(Figure 2c) or weight percent nitrogen among the time
periods. There are also no significant differences among the
radial variability of δ15N values in the white oak from the
center of the stand during all time periods and the pre-event
values from the perimeter elm. However, the radial variability
of the elm during the post-event period is significantly greater
than the white oak during the same period (p ) 0.0012). This
indicates that radial variability in tree-ring δ15N values of
perimeter trees can occur reflecting highly localized nitrogen
cycle perturbation. Unless an investigation is specifically
looking for an effect on one side of the tree, radial samples
should be grouped prior to nitrogen isotope analysis. We did
this throughout the current study.
Response of δ15N Values of Tree-Rings to Land-Use
Changes. The change in δ15N values in the perimeter elm
corresponds to tree-clearing for housing construction (Figure
2b). To determine whether this was a general phenomenon,
we measured an additional oak tree from the same stand in
Kingston and trees from a stand in Burlington that had also
undergone tree-clearing events. The three trees sampled from
the center of the stand at the Burlington and Kingston sites
are control trees in this investigation.
The trends in δ15N values of both control and perimeter
trees at the Burlington site are similar until 1976 when the
δ15N values of the perimeter trees increase substantially
(Figure 3). The pronounced 1.5-2‰ increase in the δ15N
values of all of the perimeter trees relative to the control
trees corresponds to the year of construction of an adjacent
high school (1976), and higher δ15N values in the perimeter
trees persist for ∼12 years. Eleven of thirteen pairs of preevent analyses are within one standard deviation, and, of the
two that are not, only one is significantly different (1960, p
) 0.026).
We observed a similar trend at the Kingston site. The
trend in δ15N values from the Kingston perimeter oak tree
is similar to the control trees until 1990 when they diverge
from one another (Figure 4). This corresponds precisely with
the timing of adjacent clear-cutting and land-use change.
The δ15N values of the perimeter tree remain high relative
FIGURE 3. δ15N values of red oak tree-rings from the Burlington
site. The filled diamonds are the average of three red oaks from
the center of the stand. Empty squares represent the average of
three red oaks from the perimeter. Error bars indicate one standard
deviation. Time has been divided into pre-event, post-event, and
recovered intervals. Asterisks above data points indicate the p-value
of t-tests (*p < 0.05, **p < 0.01, ***p < 0.005).
FIGURE 4. δ15N values of white oak trees from the Kingston site.
The filled diamonds are the average of three white oaks from the
center of the stand. Empty squares represent the white oak from
the perimeter. Error bars indicate one standard deviation on the
average. Analytical error on the single tree is (0.3‰. Time has
been divided into pre-event and post-event intervals. Asterisks above
data points indicate the p-value of z-tests (*p < 0.05, **p < 0.001).
to the control trees until the tree is felled by the ice storm
(February 1998). All but three of 18 pairs of pre-event (195488) δ15N values are within error and are significantly different
(1966, p < 0.0126; 1970, p ) 0.0202; 1984, p ) 0.0345).
The post-event shift in δ15N values at both sites was
compared to establish whether tree-clearing and land-use
change events at both sites resulted in similar magnitude
shifts in nitrogen isotopic composition. To remove any longterm trends present in the perimeter and background trees
from the effects of tree-clearing and land-use change, the
VOL. 39, NO. 20, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7781
difference between the average perimeter δ15N value and
the average control δ15N value was calculated for each date
to determine control tree subtracted δ15N values. The average
control subtracted post-event δ15N values at both the
Burlington and the Kingston sites were not different from
one another, but were significantly different from the preevent and recovery periods at both sites (p < 0.0001).
To examine whether there were growth-related nitrogen
isotopic fractionations, tree growth rates as indicated by
annual basal area increments (BAI) from perimeter trees
before and after the tree-clearing event were compared to
stand center trees. Specifically, the BAI from perimeter trees
during the post-event interval of elevated δ15N values were
divided by the average BAI of center stand trees for each
post-event year. This ratio was averaged for the entire period
and compared to the average ratio for a similar length time
interval immediately prior to tree-clearing. There was no
significant relative increase in growth rate of perimeter trees
after tree-clearing at the Burlington site. Both the perimeter
oak and the perimeter elm had significant increases in relative
growth rate after their respective tree-clearing events (oak,
p ) 0.0065; elm, p < 0.0001). The lack of consistent significant
changes in growth rate after the tree-clearing and land-use
changes suggests that growth rate is not responsible for the
shifts in δ15N values of tree-rings. For these samples, the
shift in δ15N values of tree-rings is more indicative of nitrogen
cycle perturbation than are changes in growth rate.
Substantial tree-removal correlates with a significant
increase in δ15N values of perimeter trees, as shown at both
the Burlington and the Kingston sites (Figures 3 and 4).
Alteration of the nitrogen cycle results in the 1.5-2‰ increase
in δ15N values of tree-rings in perimeter trees relative to
control trees at the Burlington site and the 2.5‰ increase in
δ15N values at the Kingston site. At both sites, the increase
in the δ15N values of perimeter trees occurs immediately
after the adjacent portion of the forest is removed. The
immediate shift to higher δ15N values observed following
clear-cutting in the rings of both red and white oak perimeter
trees is consistent with previous investigations wherein δ15N
values of foliage increased by 3‰ (13), and that of grass
increased by 4‰ after clearing (12). It has also been
demonstrated that the δ15N value of soils can reflect nitrate
loss and nitrogen mineralization in a similar manner (27).
Relation between δ15N Values and the Local Nitrogen
Cycle. The perimeter trees at the Burlington site took ∼12
years to recover to the same values as the control trees.
Although the relative increase in δ15N values of the Burlington
perimeter trees may be caused by input of exogenous nitrogen
following land-use change, the magnitude of the shift and
the eventual return to pre-event values is consistent with
previous clear-cutting studies and argues for a similar
disruption in the nitrogen-cycle causing the shift. A new
equilibrium of the nitrogen cycle at the perimeter and center
of the forest appears to have eventually been established.
The Kingston perimeter oak tree was felled before the treering δ15N values recovered to pre-event values.
The extraction experiment results indicate that the
transition of sapwood to heartwood removed all but the least
labile nitrogen compounds. This suggests that the δ15N value
of a growth ring is the average of the δ15N value of nitrogen
compounds taken-up by the tree in that year and the δ15N
values of the labile nitrogen compounds in all of the sapwood.
In addition to potentially dampening the initial signal, this
will smear the signal throughout the sapwood. This is most
likely responsible for the δ15N values in the perimeter trees
being elevated for ∼12 years after tree-clearing, rather than
4-9 years as observed in foliage (13) and grass studies (12).
In the perimeter trees, the length of time with elevated δ15N
values most likely reflects the length of time the bioavailable
nitrogen had an elevated δ15N value plus the residence time
7782
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 20, 2005
of nitrogen in the tree. The timing of the event is faithfully
recorded, although the amplitude and total duration of the
effect may be affected by the amount of sapwood present in
the tree. A similar effect was seen in the tracer study by Hart
and Classen (2003) (18).
Highly localized effects on the nitrogen cycle due to
perturbations are supported by the lack of δ15N value response
to tree-removal in trees from the center of the stand at the
Burlington and Kingston sites. The results of the radial
variability experiment, specifically the smaller increase in
δ15N values of the core taken from the side of the tree facing
the remaining stand (Figure 2), indicate that indeed the effect
can be localized to one side of the tree. These results suggest
that trees must be immediately adjacent to the perturbation
to the nitrogen cycle to record the effect in their tree-ring
δ15N values.
Increased rates of nitrification and nitrate loss have been
shown to cause increases in δ15N values of foliage of a
magnitude similar to that of the shift in tree-ring δ15N values
seen in the perimeter trees. In a series of nitrogen fertilizer
studies in forest ecosystems, Högberg (1990; 1991) (8, 28)
and Högberg et al. (1992) (9) argued that increased rates of
nitrification enriched the remaining ammonium available
for plant uptake. Higher foliar δ15N values have also been
observed in unfertilized watersheds experiencing high nitrate
losses (5-7, 29, 30), an effect consistent with the trend in
δ15N values of tree-rings from perimeter trees in this
investigation. This suggests that similar nitrogen cycle
perturbations are responsible for the shifts in δ15N values
observed in this investigation.
Numerous changes in the nitrogen cycle may be reflected
in the δ15N values of tree-rings. Addition of extraneous
nitrogen with a different δ15N value, an increase in the rate
of any nitrogen transformation (Figure 1), changes in
composition of the microbial community or in the composition of forest vegetation, changes in regional hydrology, or
changes in the soil pH can all affect the δ15N value of nitrogen
available to the tree. These would only appear distinct from
high-frequency variability if they were persistent or chronic,
for example, if any of those factors were changed permanently. However, if multiple stressors were acting in different
directions on the δ15N value of bioavailable nitrogen, the
effect on the δ15N value of tree-rings would be unpredictable.
In this study, we have shown that red oak, white oak, and
slippery elm trees in the immediate vicinity of areas that
have undergone deforestation and land-use change record
the event as increases in the δ15N values of their annual growth
rings. Increased nitrification, nitrate leaching from the soil,
and denitrification associated with clear-cutting have been
shown to cause an increase in δ15N values in foliage consistent
with the increase in δ15N values recorded by the tree-rings
in perimeter trees. At both the Burlington and the Kingston
sites where the land-use was permanently changed, additional factors such as probable fertilizer use, change in
local hydrology, or nitrate use from a different soil depth
could also be affecting the nitrogen cycle. Nonetheless, the
faithful recording of tree-clearing events by shifts in the δ15N
values of tree-rings from perimeter trees demonstrates that
variations of the natural abundance ratio of nitrogen in treerings can make suitable sentinels of local perturbations to
the soil nitrogen cycle.
Acknowledgments
We are grateful to April Vuletich and Kerry Klassen in the
Queen’s Facility for Isotope Research for their analytical and
technical expertise. We would like to thank Scott Lamoureux
and the Environmental Extremes and Variability (EVEX)
Laboratory for access to and assistance with equipment for
ring width measurement. This project was supported by OGS
and Queen’s University Scholarships to A.R.B., funding and
support from NSERC Discovery and MFA grants, and
Canadian Foundation for Innovation and Ontario Innovation
Trust grants to T.K.K. This paper greatly benefited from the
constructive reviews received.
Literature Cited
(1) Asner, G. P.; Seastedt, T. R.; Townsend, A. R. The decoupling
of terrestrial carbon and nitrogen cycles. BioScience 1997, 47,
226-234.
(2) Aber, J. D.; Ollinger, S. V.; Driscoll, C. T.; Likens, G. E.; Holmes,
R. T.; Freuder, R. J.; Goodale, C. L. Inorganic nitrogen losses
from a forested ecosystem in response to physical, chemical,
biotic, and climatic perturbations. Ecosystems 2002, 5, 648658.
(3) Högberg, P. Tansley Review No. 95-15N natural abundance in
soil-plant systems. New Phytol. 1997, 137, 179-203.
(4) Kendall, C. Tracing nitrogen sources and cycling in catchments.
In Isotope Tracers in Catchment Hydrology; Kendall, C., McDonnell, J. J., Eds.; Elsevier Science: New York, 1998; pp 519576.
(5) Garten, C. T., Jr.; Van Miegroet, H. Relationships between soil
nitrogen dynamics and natural 15N abundance in plant foliage
from Great Smoky Mountains National Park. Can. J. For. Res.
1994, 24, 1636-1645.
(6) Näsholm, T.; Nordin, A.; Edfast, A.-B.; Högberg, P. Identification
of coniferous forests with incipient nitrogen saturation through
analysis of arginine and nitrogen-15 abundance of trees. J.
Environ. Qual. 1997, 26, 302-309.
(7) Mitchell, M. J.; Driscoll, C. T.; Owen, J. S.; Schaefer, D.; Michener,
R.; Raynal, D. J. Nitrogen biogeochemistry of three hardwood
ecosystems in the Adirondack Region of New York. Biogeochemistry 2001, 56, 93-133.
(8) Högberg, P. Development of 15N enrichment in a nitrogenfertilized forest soil-plant system. Soil Biol. Biochem. 1991, 23,
335-338.
(9) Högberg, P.; Tamm, C.-O.; Högberg, M. Variations in 15N
abundance in a forest fertilization trial: Critical loads of N, N
saturation, contamination and effects of revitilization fertilization. Plant Soil 1992, 142, 211-219.
(10) Högberg, P.; Johannisson, C. 15N abundance of forest is correlated
with losses of nitrogen. Plant Soil 1993, 157, 147-150.
(11) Johannisson, C.; Högberg, P. 15N abundance of soils and plants
along an experimentally induced forest nitrogen supply gradient.
Oecologia 1994, 97, 322-325.
(12) Högbom, L.; Nilsson, U.; Örlander, G. Nitrate dynamics after
clear felling monitored by in vivo nitrate reductase activity (NRA)
and natural 15N abundance of Deschampsia flexuosa (L.) Trin.
For. Ecol. Manage. 2002, 160, 273-280.
(13) Pardo, L. H.; Hemond, H. F.; Montoya, J. P.; Fahey, T. J.; Siccama,
T. G. Response of the natural abundance of 15N in forest soils
and foliage to high nitrate loss following clear-cutting. Can. J.
For. Res. 2002, 32, 1126-1136.
(14) Stewart, G. R.; Aidar, M. P. M.; Joly, C. A.; Schmidt, S. Impact
of point source pollution on nitrogen isotope signatures (δ15N)
of vegetation in SE Brazil. Oecologia 2002, 131, 468-472.
(15) Helfield, J. M.; Naiman, R. J. Salmon and alder as nitrogen sources
to riparian forests in a boreal Alaskan watershed. Oecologia
2002, 133, 573-582.
(16) Poulson, S. R.; Chamberlain, C. P.; Friedland, A. J. Nitrogen
isotope variation of tree rings as a potential indicator of
environmental change. Chem. Geol. (Isot. Geosci. Sect.) 1995,
125, 307-315.
(17) Peñuelas, J.; Estiarte, M. Trends in plant carbon concentration
and plant demand for N throughout this century. Oecologia
1997, 109, 69-73.
(18) Hart, S. C.; Classen, A. T. Potential for assessing long-term
dynamics in soil nitrogen availability from variations in δ15N of
tree rings. Isot. Environ. Health Stud. 2003, 39, 15-28.
(19) Saurer, M.; Cherubini, P.; Ammann, M.; De Cinti, B.; Siegwolf,
R. First detection of nitrogen from NOx in tree rings: a 15N/14N
study near a motorway. Atmos. Environ. 2004, 38, 2779-2787.
(20) Duggin, J. A.; Voigt, G. K.; Bormann, F. H. Autotrophic and
heterotrophic nitrification in response to clear-cutting northern
hardwood forest. Soil Biol. Biochem. 1991, 23, 779-787.
(21) Nadelhoffer, K. J.; Fry, B. Nitrogen isotope studies in forest
ecosystems. In Stable Isotopes in Ecology and Environmental
Science; Lajtha, K., Michener, R. M., Eds.; Blackwell Scientific
Publishers: Oxford, 1994; pp 22-44.
(22) Faure, G. Principles of isotope geology, 2nd ed.; John Wiley &
Sons: New York, 1986.
(23) Kozlowski, T. T.; Pallardy, S. G. Physiology of Woody Plants, 2nd
ed.; Academic Press: San Diego, CA, 1997.
(24) Sheppard, P. R.; Topa, M. A. Physical-chemical pretreatment of
wood for measuring tree-ring nitrogen. 6th International
Conference on Dendrochronology; Quebec City, Canada, 2002;
pp 303-304.
(25) Sheppard, P. R.; Thompson, T. L. Effect of extraction pretreatment on radial variation of nitrogen concentration in tree rings.
J. Environ. Qual. 2000, 29, 2037-2042.
(26) Merrill, W.; Cowling, E. B. Role of nitrogen in wood deterioration: amounts and distribution of nitrogen in tree stems. Can.
J. Bot. 1966, 44, 1555-1580.
(27) Vervaet, H.; Boeckx, P.; Unamuno, V.; Van Cleemput, O.;
Hofman, G. Can δ15N profiles in forest soils predict NO3- loss
and net N mineralization rates? Biol. Fertil. Soils 2002, 36, 143150.
(28) Högberg, P. Forests losing large quantities of nitrogen have
elevated 15N:14N ratios. Oecologia 1990, 84, 229-231.
(29) Garten, C. T., Jr. Variation in foliar 15N abundance and the
availability of soil nitrogen on Walker Branch Watershed. Ecology
1993, 74, 2098-2113.
(30) Nohrstedt, H.-Ö.; Sikström, U.; Ring, E.; Nasholm, T.; Högberg,
P.; Persson, T. Nitrate in soil water in three Norway spruce
stands in sourthwest Sweden as related to N-deposition and
soil, stand, and foliage properties. Can. J. For. Res. 1996, 26,
836-848.
Received for review April 15, 2005. Revised manuscript received July 13, 2005. Accepted August 1, 2005.
ES050733P
VOL. 39, NO. 20, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7783