Introduction
Since the industrial revolution, peatlands have been subject to
increased deposition of anthropogenic nitrogen (N), as a result of fossil
fuel burning and agricultural use . The overall
consequences of enhanced N deposition in ombrotrophic peatbogs are poorly
understood, but bogs are likely to be sensitive to enhanced N inputs, because
they are adapted to conditions of very low N availability
. When N deposition exceeds plant demand, the additional N
may be used by soil microbes, and can result in the production of the
greenhouse gas nitrous oxide (N2O) via nitrification and denitrification
. Of the total N applied to
agricultural land and arising from livestock waste which is subsequently
deposited on semi-natural land, it estimated that 1 % is re-emitted as
N2O . This so-called “indirect” emission of N2O is
a large, but uncertain, term in the national inventory of greenhouse gas
emissions. There have been relatively few experimental attempts to determine
these emissions on peatbogs, and most information comes from
Fenno-Scandinavian bogs .
N deposition may lead to changes to peatbog ecosystems which influence the
emission of N2O in complex ways, particularly via soil chemistry and
vegetation composition . For example, N
deposition may affect soil pH, which affects the composition of the microbial
community , and affects the relative prevalence of the
biochemical pathways by which denitrification produces N2 or N2O
. Sphagnum mosses can immobilise a significant
proportion of the incoming N deposition . If
sphagnum cover is reduced as a result of N deposition
, more N may become available to denitrifying microbes,
and result in greater emissions of N2O . The effects may
also depend on the form of deposited N, whether reduced N (NHx) coming
predominantly from animal waste, or oxidised N (NOx) coming from energy
combustion, and whether deposited as a gas (NH3) or in rainfall (NH4+
or NO3-).
This paper reports measurements of N2O emissions, as part of a long-term
experiment in which additional N has been deposited on a peatbog in central
Scotland, for over 13 years, in three different forms (as NH3 gas, as
NH4+ solution or NO3- solution). The automated experiment was
designed to provide realistic N deposition, in terms of doses, frequency
(>100 spray events yr-1), and exposure concentrations, reflecting the
pollution climate experienced in the UK. Ambient N inputs at the site are
low, so that the responses should be representative of the more pristine
northern European peatbogs. Previous results from the experiment have
demonstrated that high doses of NH3 reduces the cover of several plant
species, but that the effects of NH4+ and NO3- on vegetation
composition and cover are not large
. Here, we examine the effects
of the dose and form of N deposition on emissions of N2O. Preliminary
data
on N2O fluxes were reported by , showing an increase
with NH3, but no effect of NH4+ and NO3-. Here, we analyse an
additional 5 years of data collected at a wider range of locations, and with
further time for any treatment effects to accumulate. The aims were to
investigate (i) the extent to which N2O emissions are stimulated by N
deposition, and whether the 1 % emission factor used in IPCC inventories
is accurate, (ii) whether the form of N deposition is important, and
(iii) whether other changes induced by N deposition (e.g. on soil chemistry
or vegetation) have an indirect effect on N2O emissions.
Materials and methods
Field site
Whim bog in the Scottish Borders (3∘16′ W, 55∘46′ N) represents
a transition between a lowland raised bog and blanket bog, on 3–6 m of deep
peat. Mean temperatures of the air and soil (at 10 cm depth) were 8.6
and 7.7 ∘C respectively (2003–2009 means). The annual rainfall was
1092 mm (734–1462 mm range). On average, the water table was 10 cm below
the surface of the peat in the hollows, i.e. relatively wet for most of the
year. Hummocks were typically 20 cm higher than the hollows. The peat was
very acidic, with pH 3.4 (3.27–3.91 in water). The vegetation was classified
as a Calluna vulgaris–Eriophorum vaginatum blanket mire community
(M19 in the UK National Vegetation Classification, Rodwell, 1998). Replicate plots were highly variable and dominated by unmanaged
Calluna of variable age and stature occurring as mosaics containing
Calluna vulgaris and Sphagnum capillifolium hummocks and
hollows containing S. fallax and S. papillosum. Other
common species included Erica tetralix and the mosses Hypnum jutlandicum and Pleurozium schreberi.
Experimental Treatments
Nitrogen was applied to the site using two different treatment systems, for
dry deposition of NH3 gas, and wet deposition of NH4+ and NO3- in
solution. Treatments commenced in June 2002 and continued all year round,
except when temperatures were near freezing.
Spatial distribution of N deposition in the Whim experiment. The
ammonia release system was located at the centre of the boardwalk structure
in the middle of the image. Ammonia deposition was calculated from
concentration measurements downwind of the ammonia source using the method of
, and interpolated across the central rectangular area of the
image using ordinary kriging. In the wet deposition plots (circles in the
lower part of the image), the nitrogen solution applied was sprayed evenly
across the whole 12.8 m2 plot. North is at the top of the image.
![]()
NH3 deposition was manipulated using a free-air release system
. NH3 was supplied from a cylinder of pure liquid NH3,
diluted with ambient air and released from a perforated 10 m long pipe, 1 m
off the ground. NH3 was released only when the wind direction was in the
south-west, between 180 and 215∘, temperatures exceeded freezing, and
wind speed exceeded 2.5 m s-1. This produced a sector downwind wherein
NH3 decreased with distance from the fumigation source. NH3
concentrations were measured 0.1 m above the vegetation using passive ALPHA
samplers at 8, 12, 16, 20, 24, 32, 48, and 60 m from the
source along the transect. A detailed profile was measured to capture the
concentration gradients both vertically and horizontally .
Ammonia deposition was calculated from the concentration measurements, using
the method of . The deposition at these locations was
interpolated using ordinary kriging, as shown in
Fig. , assuming the deposition velocity was
spatially homogeneous.
Wet deposition of NH4+ and NO3- was experimentally increased in a
number of replicated plots in a randomised block design, using a water
sprayer system . Concentrated solutions of either
NH4Cl or NaNO3 were diluted in rainwater and transferred to each plot
via 100 m lengths of 16 mm pipe. Each pipe terminated in a central sprayer
with a 360∘ spinning disc that distributed the solution uniformly over
the 12.8 m2 plot. The volume of solution applied to each plot was
monitored using a water meter on each supply line. Three treatment levels
were applied, aiming to provide total N deposition rates of 16, 32, and
64 kg N ha-1 yr-1, in addition to a control treatment which only received
ambient N deposition (8 kg N ha-1 yr-1). The three treatment levels
were achieved by applying either NH4Cl or NaNO3 solution at
concentrations of 0.57, 1.71, or 4.0 mM. Wet treatments increased
precipitation amounts by ca. 10 %. Control plots receive the additional
rainwater without any additional nitrogen. There were four blocks, with one
treatment level in each, to give a total of 28 plots. The sprayer system was
automatically triggered every 15 min (so long as there was sufficient
rainwater in the collection tank) air temperature was above 0 ∘C, and
wind speed was above 5 m s-1. This produced a realistic pattern of high
frequency, extensive nitrogen deposition, with ca. 120 applications per year.
Daily mean nitrogen deposition rates in the NH4+ and NO3-
treatments and at 16 m downwind on the NH3 transect. NH4+ and
NO3- deposition was calculated from the duration of spraying events each
day, multiplied by the NH4+ and NO3- concentration in the solution,
and accounting for the area covered by the spray. NH3 deposition was
calculated from the duration of fumigation events each day, measurements of
NH3 concentration, and meteorological data, using the method of
.
![]()
Soil water samples were extracted from dipwells in all plots at the same time
as gas flux measurements were made. Concentrations of soil water NH4+ and
NO3- were measured by ion chromatography following filtration. The
detection limits were 0.014 and 0.062 mg L-1 for NH4+-N and
NO3--N, respectively. The percent cover of each vegetation species was
recorded within each chamber location every few years.
Greenhouse gas exchange
Nitrous oxide fluxes were measured by the static chamber method
, typically on a monthly basis. Cylindrical PVC collars
(38 cm in diameter and typically 25 cm high) were inserted into the peat
within each plot.
On each sampling occasion, a lid was sealed on top, and left in place for
30–40 min. Four 20 mL samples were removed by syringe through a 3-way tap
or rubber septum, stored in vials or Tedlar bags, and analysed on a gas
chromatograph (5890 series II, Hewlett Packard), together with replicates of
three or four standard gases with known concentrations. For each sequence of
gas samples from a chamber, the flux was calculated as:
F=dCdt0⋅ρVA,
where F is gas flux from the soil (µmol m-2 s-1),
dC/dt0 is the initial rate of change in concentration
with time in μmol mol-1 s-1, ρ is the density of air
in mol m-3, V is the volume of the chamber in m3,
and A is the ground area enclosed by the chamber in m2.
The parameter dC/dt0 was calculated using linear and
non-linear asymptotic regression methods . Using a mixture of
goodness-of-fit statistics and visual inspection, the regression method that
provided the best fit for the time series of concentration was chosen for
each individual measurement. With this method of flux calculation, any
non-linearity should be accounted for as far as possible. However, the time
resolution (approximately 10 min) limits the detectable degree of
non-linearity in the initial concentration change, so there remains some
potential for underestimation of fluxes .
Response of N2O flux to deposition of NH3, NH4+, NO3-,
and total N deposition. Error bars show the 95 % confidence interval,
based on the regression slope for each flux measurement. Solid lines show
the fitted response from the general additive mixed model. Dotted lines show
the emission predicted with the IPCC default emission factor, as 1 % of
the deposited N.
![]()
Statistical analysis
The data were first analysed using a linear mixed-effects model
. There were 729 flux measurements in total, after
removing four outlying measurements above 10 nmol m-2 s-1 and
two below -2 nmol m-2 s-1. We fitted fixed-effect terms for
soil temperature, Tsoil, water table height, zwater
(negative values indicate depth below the surface), ammonia-N deposition
rate, FN-NH3, ammonium-N deposition rate, FN-NH4,
and nitrate-N deposition rate, FN-NO3, and random-effect terms
with a design matrix Zi,j to account for the repeated measures
on each chamber location, j, nested within each experimental block,
i:
FN2O,ij=β0+β1⋅Tsoil,ij+β2⋅zwater,ij+β3⋅FNH3,ij+β4⋅FNH4,ij+β5⋅FNO3,ij+bi⋅Zi,j+bij⋅Zij+ϵijbi∼N0,σ12bij∼N0,σ22ϵij∼N0,σ32.
The data were also analysed using a general additive mixed-effects model
, with the same fixed- and random effect terms, but allowing
for non-linearity in the fixed-effect responses. To analyse the relationship
between N2O flux and vegetation species composition, we used a
multivariate approach, partial least squares regression
PLS,. The approach is an extension of principal
components analysis (PCA), but whereas PCA focuses on the variance in a
matrix of variables,
X, PLS computes the scores and loadings in
such a way to describe the covariance between X and a response
variable or matrix, Y. In this context, we have a matrix
consisting of the percent cover of each plant species in each chamber, and
the response variable is the N2O flux. PLS should perform better than PCA
in situations where an infrequent species (contributing little to the
variance in X) is highly correlated with Y. In PLS,
such a component would automatically be present in the first component, but
would be a minor component in PCA.
Results
Figure shows the spatial distribution of N
deposition in the dry and wet deposition treatments. Deposition of NH3
peaks at around 100 kg N ha-1 yr-1, just downwind of the fumigation
source. NH3 deposition decreases with downwind and cross-wind distance
from the fumigation source, and approximates a Gaussian plume pattern
expected from micrometeorological theory. The plume is aligned slightly to
the east of the boardwalk transect, although there is some uncertainty in the
interpolation between NH3 samplers. Deposition of NH4+ and NO3- on
the wet deposition plots are shown on the same colour scale. These are known
with much greater certainty, as no modelling step is required.
Figure shows the time
series of nitrogen deposition in the NH4+ and NO3- treatments, and on
the NH3 transect at 16 m, where annual deposition was similar to that in
the NH4+ and NO3- treatments. Distribution of deposition events over
time is similar in both treatments. Deposition events were spread over most
days of the year, with no deposition only in a short period in mid-winter.
NH3 deposition is calculated as a function of stomatal conductance
, so rates are higher in daytime and in the summer.
Whilst there is considerable scatter in the response of N2O fluxes to mean
annual nitrogen deposition, an increase in N2O flux with NH3 deposition
was apparent (Fig. ). No trend with NH4+ or
NO3- deposition was obvious. In many of the flux measurements, the
magnitude of N2O fluxes was close to the measurement error in the static
chamber method. In both the NH4+ and NO3- treatments, only 9 % of
fluxes (respectively) had 95 % confidence limits which did not include zero.
Detecting a clear response is inevitably difficult when measurement noise
contributes substantially to the variability in the data. By contrast, in the
NH3 treatment, 40 % of fluxes had confidence limits which did not include
zero.
Fitted response of N2O flux to soil temperature, water table
height, and deposition of NH3, NH4+, and NO3-, as estimated by the
general additive mixed model. Partial residuals are shown for the smooth
terms, which are the residuals that would be obtained by dropping the term
concerned from the model, while leaving all other estimates fixed. NH4+
and NO3- were treated as linear terms. Upper and lower lines show 2
standard errors above and below the fitted estimate.
![]()
The output from the linear mixed model analysis is shown in
Table , with the coefficients representing the response to the fixed factors.
As well as showing significant responses to temperature and water table depth,
N2O fluxes responded to NH3 deposition (Fig. ).
This response was greater than the default 1 % IPCC emission factor, and
comes close to 8.5 % (with the appropriate unit conversion).
The relationship may not be linear , and the
general additive mixed model (GAMM) was fitted to allow for non-linearity
in the fixed effects. However, the exact form of the response to NH3
deposition was not well constrained by the data, especially at the lower
values, and a simple linear fit was justified (Fig. ).
The modelled effect of NH4+ and NO3- deposition was slightly negative on
average, although positive and negative slopes are both plausible (Table , Fig. ).
NH4+ concentrations in the soil water were elevated in the high NH4+
deposition treatment, by around 1 mg N dm-3 on average
(Fig. ). By contrast, the high
NO3- deposition treatment had no clear effect on NO3- concentrations
in the soil water. On the NH3 deposition transect, there was a clear trend
in soil water concentrations of both NH4+ and NO3- with NH3
deposition (Fig. , right-hand
plots). At the equivalent level of NH3 deposition, NH4+ concentrations
in the soil water were elevated by a similar amount to that in the high
NH4+ deposition treatment. There were weak relationships between N2O
flux and NH4+ and NO3- concentrations in the dry treatment, but no
clear relationship in the wet treatment.
Results of fitting the linear mixed-effects model (Eq. 2) to the
data by maximising the restricted log-likelihood. Columns show the
denominator degrees of freedom, F values, p values from Wald tests for
each term, and the β coefficients in Eq. (2).
DF
F value
p value
Coefficient
Intercept
392
23.90
0.000
-0.1664
Tsoil
392
16.75
0.000
0.0181
zwater
392
22.81
0.000
0.0528
FNH3
40
29.96
0.000
0.0092
FNH4
40
1.27
0.267
-0.0046
FNO3
40
1.74
0.194
-0.0041
NH4+ and NO3- concentrations in the soil water in the
experimental treatments. In the right-hand plots, colours show measurements
grouped by distances downwind of the fumigation source.
![]()
Mean N2O flux at each chamber location fitted by the PLS model
based on vegetation species composition plotted against observed mean fluxes.
The data are grouped by nitrogen treatment form (symbol shape) and dose
(symbol colour). The solid line shows the 1 : 1 line.
![]()
Vegetation composition provided reasonably good explanatory power for N2O
flux, and the PLS regression explained 56 % of the variance in N2O flux
(Fig. ). The first two components explained
27 % of the variance, and gave an interpretable ordination of the chambers
(Fig. ). The chambers high on the first axis were
dominated by Eriophorum vaginatum, often damaged by NH3, with
little or no moss cover, and had high fluxes. The chambers low on the first
axis had high cover of Sphagnum capillifolium and low fluxes.
The second axis differentiates hummock and hollow vegetation, and a soil
moisture difference. The drier hummocks with Calluna vulgaris and
Deschampsia flexuosa had lower fluxes than the hollows.
Biplot showing the scores for each chamber location for the first
two PLS components. The first two PLS components represent the orthogonal
indices of vegetation species composition which are the best linear
predictors of mean N2O flux at each chamber location. The loadings for the
important plant species are superimposed, indicating which species contribute
most to the components. The data are grouped by nitrogen treatment form
(symbol shape) and dose (symbol colour); symbol size is proportional to the
magnitude of the mean N2O flux.
![]()
Discussion
Our results confirm the early findings of that there
was no clear response of N2O flux to deposition of NH4+ or NO3-,
whereas high doses of NH3 reduced the cover of Calluna vulgaris
and Sphagnum species, and increased N2O flux. Other results in
the literature show a range of responses of N2O emission to experimental N
addition, from no response to substantial increases. found
no effect of experimental N addition (NH4NO3) on N2O emissions from
two Swedish bogs, and peak fluxes were less than 1 nmol m-2 s-1,
when 40 kg N ha-1 yr-1 was applied in only three relatively
large doses. found no response of N2O emission to
additions of up to 100 kg N ha-1 yr-1 (NH4NO3) to a
Sphagnum fuscum pine bog in Finland, over a 6-year study. Following
nitrate addition to ex situ peat cores from Polish sedge fen,
observed no increase in N2O emissions from cores
from vegetated tussocks or unvegetated hollows, except for an increase of
0.15 nmol m-2 s-1 at their low nitrate (KNO3) addition rate.
Some clearer positive responses have been observed where bogs have been
drained, or where very high levels of nitrogen have been applied.
found that N2O emissions were increased by up to 0.8 nmol m-2 s-1 for around nine months after a single experimental N
addition of 100 kg N ha-1 yr-1 on a drained and forested peatland in
Finland; KNO3, NH4Cl, and urea gave a similar range of responses.
observed an increase of 0.3 nmol m-2 s-1 with
the application of 240 kg N ha-1 yr-1 (NH4NO3) to a freshwater
marsh in China, bi-weekly over the summer growing season. Clear responses
can, however, be very short-lived. For example, found
a short-term response of N2O efflux to NH4NO3 addition in an
in vivo study of soil from an alpine peatland in Tibet, but
differences from the control lasted less than 10 days.
A response of N2O emissions to nitrogen addition is more often detectable
in laboratory incubations, where there are fewer feedbacks and interactions.
Field studies commonly show complex interactions with other variables,
resulting in no effect in the field, or making interpretation of results
complicated. In a Finnish spruce swamp buffer zone, found
that nitrogen addition increased N2O efflux in laboratory incubations, but
in situ N2O effluxes were low and unresponsive.
found that N2O fluxes were positively correlated with the numbers of
nitrite oxidisers, nitrification potential, N, P, Ca, and pH of the soil
and negatively correlated with the level of the water table and K content of the soil. In a
study by on a restored peatland in Finland, N2O
emissions showed an asymptotic increase with nitrate concentration, and an
exponential decrease with E. vaginatum cover. The interpretation was
that N2O emission was the outcome of resource competition for nitrate
between denitrifying bacteria and (E. vaginatum) roots, but both
have limited capacities for uptake of nitrate.
We can interpret our results similarly in terms of resource competition for
nitrate. The wet deposition of NO3- did not increase concentrations of
NO3- in the soil water. Previous data suggest that all the additional N
deposited on the wet treatment plots accumulated in the top 10 cm of the peat
and below-ground vegetation (based on the accumulation of N in the different
pools shown in , Fig. 7, although estimates are rather
uncertain because of sensitivity to errors in bulk density, C and N
concentrations, and sampling error). The deposited nitrogen was thereby
immobilised, and not available to soil microbes, and was not denitrified to
N2O. The wet deposition of NH4+ did increase concentrations of
NH4+ in the soil water, but NH4+ is less directly related to the
microbial production of N2O. In the dry deposition treatment at high
levels of NH3 deposition, although there was similar accumulation of the
deposited N in the peat, much of the vegetation cover was killed, leaving
only a sparse cover dominated by E. vaginatum. The deposited N was
therefore more available to soil microbes because there was less competition
with plants, and could be denitrified to N2O.
One of the difficulties in interpreting these results is that there are
multiple microbial processes occurring (nitrification, denitrification, and
N2O consumption), but the chamber flux measurements only give the net
exchange of N2O. The treatments could affect these multiple processes in
different ways, e.g. via their effects on pH (and thereby influencing the
microbial populations), or more directly by altering the amount of substrate
for denitrification. However, without detailed information on the
nitrification and denitrification potentials, it is difficult to draw
stronger conclusions. Some studies have suggested that gross uptake of N2O
can be substantial, at least in forest soils
. In our study, the negative fluxes are
so small compared to the measurement error, that no real pattern can be
discerned, and this is a common finding in agricultural systems
.
The multivariate analysis of the vegetation provided the best means for
explaining the variation in N2O fluxes. Shifts in the species composition
appeared to be the clearest sign that N deposition was affecting the
long-term outcome of resource competition for nitrate, and thereby
influencing N2O flux. A similar result was found in a multivariate
analysis of vegetation in relation to methane flux in the UK ecosystems
. Direct relationships between N2O flux and both
nitrate and ammonium were poor, possibly because these are influenced by
feedback from the flux itself.
The unique aspect of the experimental design, in providing a very realistic
simulation of nitrogen deposition rates, in terms of frequency and exposure
concentrations, makes detecting effects more difficult. Unlike measuring
N2O fluxes on agricultural land, where large peak emissions typically
follow fertiliser applications, here we are measuring small, diffuse N2O
fluxes spread over the whole year, which is necessarily more difficult.
Experimental designs of the wet and dry deposition experiments were quite
different, because of the logistics of applying gases and liquids to
ecosystems. This may partly explain why the effect of NH3 was easier to
detect. NH3 was applied on a larger spatial scale, with very high rates of
deposition adjacent to the fumigation outlet. Implicitly, we assume that
differences with distance are effects of NH3, as there was no true
replication (i.e. there was only one transect). Given this assumption, we
take n to be 60 in the NH3 experiment. The probability of detecting an
effect size of +0.068 nmol m-2 s-1 (=1 % of 60 kg N ha-1 yr-1)
is reasonably high (p = 0.96), from a standard power analysis
, given the typical measurement error in static chamber data
(and ignoring between-plot variability). In the case of the wet deposition
experiment, there were true replicate plots in a randomised block design, but
sample size was small (4 plots per treatment). The probability of detecting
the same effect size with n=4 is rather low (p = 0.13).
In our experimental results, neither NH4+ nor NO3- deposition
increased fluxes of N2O (indeed the mean effect of both was negative), and
all the deposited N appeared to be retained in the peat and below-ground
vegetation. However, we need to be careful before concluding that the true
effect was zero (or negative), given that the effect size we were looking for
was small. An emission factor of 1 % is not inconsistent with the data,
given the between-plot variability, measurement precision and small sample
size. If this were the true emission factor, we can estimate the N2O
emission resulting from N deposition over peatlands in the UK. Combined
deposition rates of NH4+ and NO3- are generally less than 10 kg N ha-1 yr-1 on UK peatbogs
. With a 1 % emission
factor, this would imply a mean emission of 11 pmol m-2 s-1 or
1.3 g C-CO2 eq. m-2 yr-1. (To express N2O flux in
units of C-CO2 equivalents, we convert to a mass of N2O and
multiply by 298 the global warming potential of N2O over a 100 year
time period including carbon cycle feedbacks, and report only
the mass of carbon in the CO2 equivalents, i.e. 12/44.) This compares with
contemporary net ecosystem carbon balance measurements in the range of
28–50 g C-CO2 eq. m-2 yr-1 in UK peatlands
.
NH3 has a different spatial distribution, being deposited much closer to
its sources, so with high “hotspots”, but with lower overall mean
deposition rates (<5 kg N ha-1 yr-1). In the immediate vicinity of a
large agricultural source, high mean emissions of N2O would be predicted,
in the order of 0.58 nmol m-2 s-1 or 65 g C-CO2 eq. m-2 yr-1
(i.e. 8.5 % of 60 kg N ha-1 yr-1), but over relatively small areas. This approaches the
large values observed by in Arctic bare peat circles
(averaging 0.86 nmol m-2 s-1 or 97 g C-CO2 eq. m-2 yr-1), but upscaling these
emissions to national scale is challenging.