ArticlePDF AvailableLiterature Review

Nitrogen Critical Loads: Critical Reflections on Past Experiments, Ecological Endpoints, and Uncertainties



Nitrogen Critical Loads (NCL), as purported ecological dose-response outcomes for nitrogen deposition from anthropogenic sources, play a central role in environmental policies around the world. In the Netherlands, these NCL are used to assess, via calculations using the model AERIUS, to what extent NCL are exceeded for different habitats as a result of different sources such as industry, agriculture, traffic. NCL are, however, not well defined, and are subject to hitherto unrecognized forms of uncertainty. We will address this with reference to a number of key studies that forms the basis for several NCL. We will subsequently propose amendments that could be applicable to future nitrogen studies and their enhanced relevancy in decision making.
An International Journal
January-March 2022:110
© The Author(s) 2022
Article reuse guidelines:
DOI: 10.1177/15593258221075513
Nitrogen Critical Loads: Critical Reections
on Past Experiments, Ecological Endpoints,
and Uncertainties
William M. Briggs
and Jaap C. Hanekamp
Nitrogen Critical Loads (NCL), as purported ecological dose-response outcomes for nitrogen deposition from anthropogenic
sources, play a central role in environmental policies around the world. In the Netherlands, these NCL are used to assess, via
calculations using the model AERIUS, to what extent NCL are exceeded for different habitats as a result of different sources
such as industry, agriculture, trafc. NCL are, however, not well dened, and are subject to hitherto unrecognized forms of
uncertainty. We will address this with reference to a number of key studies that forms the basis for several NCL. We will
subsequently propose amendments that could be applicable to future nitrogen studies and their enhanced relevancy in decision
critical loads, nitrogen, uncertainty
Acritical load (CL)is an ofcial level of exposure to a
substance above which environmental harm is said is likely
occur. These loads are mostly presented as atmospheric de-
position rates of kilograms per hectare per year.
Nitrogen critical loads (NCL) have been at the forefront of
governmental ecological protection policies in farming prac-
tices, industrial activities, urbanization, trafc, and so on. Their
history dates back to at least the 1980s, when the rst tentative
experiments and observational studies were published on the
effects of multiple air pollutants on the natural environment.
In this contribution, we will revisit a sampling of studies
relied upon to set NCLs applied to different ecological end-
points as they are applied mainly in the Netherlands and
Europe. We will analyze these studies from an informational
standpoint; that is, we shall scrutinize the endpoints, the
experimental set-ups, and inherent uncertainties either re-
ported and applied or not. Before we can delve into the
material, however, we rst need to dene what NCL are.
The following denition for NCL is from
The term
critical load for nitrogen depositionmeans: the limit above
which there is a risk that the quality of the habitat will sig-
nicantly be affected by the acidifying and/or eutrophication
inuence of atmospheric nitrogen deposition.This is
somewhat loose, as the individual terms are undened, leaving
much room for differing interpretations.
De Vries et al,
in their Critical Loads and Dynamic Risk
Assessments Nitrogen, Acidity and Metals in Terrestrial and
Aquatic Ecosystems, dene CL, amongst which they include
NCL, as follows: Following the denition of a critical load by
Nilsson and Grennfelt, the sustainability of the structure and
function of an ecosystem is protected when a critical load is
not exceeded by (atmospheric) deposition of pollutants, thus
avoiding adverse effects and possibly irreversible damage in
the future.The extent of adverse effectis left undened.
The study referred to in this quote is the 1988-report CL for
Sulphur and Nitrogen edited by Nilsson and Grennfelt.
Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons
Attribution-NonCommercial 4.0 License ( which permits non-commercial use,
reproduction and distribution of the work without further permission provided the original work is attributed as specied on the SAGE
and Open Access pages (
Independent Researcher, Detroit, MA, USA
University College Roosevelt, Middelburg, Netherlands
Received 29 November 2021; received revised 4 January 2022; accepted 4
January 2022
Corresponding Author:
Jaap C. Hanekamp, Environmental Health Sciences, University of
Massachusetts, Amherst, MA, USA.
Email: and
report proposed the following CL-denition: A quantitative
estimate of an exposure to one or more pollutants below which
signicant harmful effects on specied sensitive elements of
the environment do not occur according to present
The CL-denition as found in Nilsson and Grennfelt
still referred to as more or less authoritative. Be that as it may,
these denitions are rather general in scope and need at least
some specication with respect to endpoints and ecologies
(types of ecosystems). The former are mostly described as
eutrophication, acidication, and pressures on biodiversity
(species richness). The latter refers to marine habitats (EUNIS
class A), coastal habitats (EUNIS class B), inland surface
water habitats (EUNIS class C), and so on. EUNIS stands for
European Nature Information System.
Importantly, the endpoints of eutrophication, acidication,
and pressures on biodiversity (species richness) interact with
each other, according to the literature, in multiple loops and
In order to properly understand NCL as actual measureable
and decisional numbers of kg/ha/yr, we propose a reverse-
engineering strategy. That means that we take the actual NCL
stated or used for some ecosystem and backtrack its origin
from the available scientic literature.
Intriguingly enough, in the Netherlands NCL are reported
as unique singular values,used ostensibly to determine
deterioration of ecosystems once N-deposition, calculations
made by AERIUS, exceeds these singular values. These
unique singular values are portrayed as the best-estimate of the
scientic state-of-art, for example, in
p. 13. About the on-line
calculation tool AERIUS, see Ref.
In one of the AERIUS
documents (AERIUS, the calculation tool of the Dutch In-
tegrated Approach to Nitrogen), N-deposition and NCL are
discussed as follows: AERIUS adds together the calculated
amount of project-related deposition and the background
deposition, and subsequently shows, per location, the total
deposition in relation to the critical load. For ecologists, this is
crucial information to assess the situation.
In the present work, we address uncertainties and over-
certainties specically, as these concepts seem undervalued in
the ecological studies we have scrutinized.
We examine what biological measures several authors have
chosen below in their effort to understand the uncertainty in
the measures themselves, and how these effect an under-
standing of the concept of critical load as such. The denitions
of critical load given here are broad and do not imply any
singular or specic or even range of policies that could or
should be taken based on any critical load.
Nitrogen Critical Load Modeling
An approach to modeling nitrogen critical loads was given in,
which we reviewed in.
Briey, in
data was gathered either
from papers detailing planned small-scale (time and space)
nitrogen-added experiments or in papers detailing large-scale
observations. The data from each paper was comprised of a
mix of plant growth, plant chemistry and nitrogen-uptake
measurements. There was no strict consistency of these
measures across the papers. Nitrogen deposition was either
scaled up from the small time and space plot experiments or
taken directly from the large-scale observations at the kg ha
level for all of the groups.
In most papers, controls, that is, no added- or low-N
groups, were compared statistically with 1 of the several
different plant growth or nitrogen-uptake measurements in N-
added experiments, or in areas with lower (control) and higher
atmospheric N. If the difference in these control-to-added-
group measures gave a wee P-value, the difference was la-
beled (in effect) harmful,else it was labeled not harmful.
The levels of nitrogen at the differences, or at the background
if there was no wee p-value, was also noted. Finally, a logistic
regression was created using these nitrogen levels and the
labels harmfulor not harmful.
If in the model an input level of nitrogen (in kg ha
gave 20% or more chance for harmful,that level of nitrogen
was called critical,The harmful, again, was not consistent
and was based upon a mixture of different outcomes, that is,
various incommensurable measures of plant growth or
Briggs and Hanekamp
provide an in-depth statistical
critique of this method, giving several suggestions for im-
provement in the modeling aspects alone. However, there are
further troubling aspects about this approach than just the
statistical modeling technical details as such.
Indeed, this approach, which was the rst attempt to dene
uncertainties as such statistically, has a number of other ob-
vious shortcomings. First, there was the mixing of planned
nitrogen-added experiments with large-scale observations.
The small plots (in time and space) were scaled up, but with no
accounting for the added uncertainty inherent in this upscaling
which must be substantial when moving from plots
measured in the small square meters and short time periods to
hectares over periods of one or more years. About this scaling,
see more below. The experiments were controlled to some
extent, but the large-scale observational measurements (typ-
ically country- or region-sized) were not (e.g.,).
The background levels of nitrogen deposition and its
variability and seasonality were not well measured and
reported across these papers. Some papers counted only
wet deposition, some only dry, and some had the total. The
uncertainty in the background levels was rarely given, or
given only crudely (e.g.,).
Single, seemingly certain
values of nitrogen were used for entire regions, with no
real idea of changes in these numbers due to seasonality or
geography (e.g.,).
As we also see and discuss below, the
biological measures differed greatly as well. This is not
inappropriate if decisions have to be made for each of these
differing measures, but there is no justication given for
combining different measurement types when deciding
2Dose-Response: An International Journal
Modeling is certainly a reasonable approach to quantify
uncertainty in well-dened critical loads, while using ho-
mogeneous data. Decisions based on model probabilities must
fully account for the costs and gains of those decisions, too, of
course. It is unlikely a one-size-ts-all value of 20% chance
for harm(dened above) to occur would be justiable in
every situation.
Here we do not attempt to critique any statistical model, or
even any statistical testing techniques used within individual
studies. Our concern comes before these modeling steps,
however necessary they might eventually be. We want to build
a consensus on the exact kinds of data needed, on the
quantiable and replicable denitions of harm, of critical
loads, and all the elements that go into models.
Our guiding discussion in this contribution is of Chapter 6 of
the major report of Bobbink and Hettelingh,
Effects of
nitrogen deposition on mire, bog and fen habitats (EUNIS
class D),which includes a wide range of wetland systems
that have their water table at or above soil or sediment level for
at least half of the year, dominated by either herbaceous or
ericoid vegetation.
Here we step through the most and paradigmatic important
papers cited and highlighted in,
discussing them with respect
to endpoints and uncertainties of the types mentioned above.
Our interest is not in the specics of the experiments outlined
per se, nor on the precise details of the plant biologies.
Moreover, we do not profess to have completed a universal
survey of this large topic. But we do believe we have attended
to the major components of uncertainty, all of which are
imperative to recognize before any modeling can commence.
Later we discuss what steps we believe need be taken
Bragazza et al
One of the most informative papers addressing the various
aspects of uncertainty is,
a work that examined ombro-
trophic Sphagnum plants in 15 mires across Europe. Ac-
cordingly, we pay this paper the most attention.
Sampling was performed per country in each of three to
six mires and three to six hummocks, all with dense cover of
Sphagnum and low cover of vascular plants. Various
Sphagnum species dominated particular mires. Each testing
site was a 10 × 10 cm plot. It is unclear how long these plots
were exposed, but the authors say sampling was done mostly
between mid-September and mid-October.”“Mostlywas not
Mean annual atmospheric measurements for precipitation,
temperature, and N- and K-deposition were taken from third-
party sources. One-number averages were used for entire regions
and across whole years. For instance, Finland was summarized
with having a background level of 0.2 g m
Sphagnum from these small plots were then analyzed,
mainly for their N to P or N to K content ratios. These
measured values did include standard error measurements,
giving some indication of uncertainty of measurements over
the 10 × 10 cm plots.
These chemical and biological measures where then input
into regressions as outcomes with predictors being the
country-or region-level single atmospheric N levels. It appears
the locations or groupings of the individual plots were not
controlled for, nor was species considered, though the dif-
ference between lawns and hummocks was noted in separate
regressions. Not controlling for these various things would
tend to, optically only, decrease any uncertainty in the re-
sultant regressions, though we cannot here tell by how much,
because no such uncertainties of the regressions were shown.
That is, the model would not account for all relevant
The authors dened CL as the amount of N bulk depo-
sition above which Sphagnum plants experience nutrient
imbalance to such an extent to greatly decrease the absorption
of exogenous N. In this sense, N deposition levels above the
critical load cause a N saturation of the Sphagnum layer, that
is, the removal of N limitation associated with a decrease of N
retention capacity. The atmospheric N input shifting ombro-
trophic Sphagnum plants from being N limited to being P + K
colimited is c. 1 g/m
, a critical load value consistent
with the value suggested by other authors.
Figure 1. Mean values (± SE) of (A) N:P and (B) N:K ratios in
hummock and lawn Sphagnum plants at each mire in relation to
atmospheric N deposition. Dashed and continuous lines represent
the theoretical patterns based on regression modelfrom.
gure is also cited in.
Briggs and Hanekamp 3
This is a curious denition. It is true that if the local
ecosystem has abundant N, then it could be P- or K-limited,
but reaching these limits does not by itself imply any critical
level has been reached. For instance, there could just be too
low levels of P or K. It also does not appear to be a problem
that plants do not store N at the same rates when N is abundant
because, of course, N is abundant. In other words, just why
these levels are critical is ambiguous.
In their report,
reference the central gure from,
included here in Figure 1. This shows the N:P and N:K ratios of
plant matter, plotted against the single-value atmospheric N
estimates. Regressions are over-plotted separately for lawns and
hummocks. Both the K and P limiting values are clear enough
when atmospheric N exceeds about 1 g m
exceeds this amount, plants have all the internal N they need,
but could possibly use more P and K if it were available.
This does not make that 1 g m
though. For if the environment had more available P and K,
then the plants could conceivably take up more, and those N:K
and N:P ratios would be pushed up to higher levels of at-
mospheric N. If anything, all this kind of picture indicates by
itself is that once atmospheric N goes beyond a certain point,
the effects of N are limited. They saturate with N, and perhaps
not store as much, but then they do not need to, because there
is plentiful atmospheric N to make up for whatever capacity to
store is lost.
Based on very small sample sizes (number of mires 5)
capitulum biomass (g dm
) was slightly less for atmospheric
N greater than 1 g m
in hummocks, but slightly
greater for lawns. The same was reported for stem volumetric
density (g dm
). The signal is thus mixed; plus, there is no
reporting of biomass or density at other values of atmospheric
N. Which was scarcely possible given the small sample. And
no other accounting of local conditions were given, so any
indication of cause (N is not the only potential cause of change
of the measurements taken) must be considered incomplete at
Berendse et al
A second representative paper is,
which describes experi-
ments in which 25 × 37.5 cm in situ plots in Finland, Sweden,
Switzerland, and the Netherlands, each supplied with either
extra CO
piped through hoses, or extra NH
added to
water and sprayed on the plots. This was said to be equivalent
of adding 3 g m
at three of sites, except the
Netherlands, where it was equivalent of 5 g m
the previous paper, single-number summaries were used for
atmospheric N, ranging from .4 g m
(a site in
Finland) to 3.9 g m
(a site in the Netherlands). These
were not actual measurement, either, but output from a model
(RAINS 7.2; see the paper for a description).
At the end of the experiment, a knitting needle was laid on
the plots and the number of species touching it were recorded.
Sphagnum dry weights and lengths were also sampled. The
added CO
and NH
experiments were compared with
backgrounds. We do not here consider the CO
Sphagnum production was greater in Finland in the added-
N experiment contrasted with ambient conditions, and was
less in other three countries. There was considerable vari-
ability across the four countries. N concentration was greater
in Sphagnum plants in the added-N experiments in all four
countries. There were more vascular plants noted in the added-
N experiments than in ambient conditions.
The above-ground biomass was greater in two countries
(Finland and Netherlands) and less in two others (Sweden and
Switzerland) in the added-N experiments. But there was less
below-ground biomass in all but the Netherlands, which had
more. There was more standing dead and litter biomass (more
carbon sequestration) in all four countries in the added-N
None of the differences in these experiments between
ambient conditions were especially large, the signals are
decidedly mixed, and as always it becomes difcult to know
how to extrapolate the very small-scale experiments to entire
regions, or even countries. It has the same difculty as the
previous paper in which the actual ambient levels of atmo-
spheric N were unknown, and only assumed.
There is no way to draw, from this study, any critical value
of N such that sufcient (or specic) undesirable changes have
taken place. A critical value affecting biology could be de-
ned, of course. Say, the percent coverage of non-Sphagnum
plants exceeding some pre-specied level. But this would
have to come with a plus-or-minus when extrapolating the
experimental values to regions or whole countries.
Tomassen et al
Another paper was,
describing an experiment in a simulated
environment. Several 24 × 24 × 32 cm jars into which extra N
was injected were used. Varying amounts of N were added,
from 0 to 4 g m
, with negligiblebackground
deposition level of N. This negligibility was likely true be-
cause the environment was wholly articial, unlike in other
experiments which dealt with atmospheric background levels.
Differences were found at the highest N for plant biomass
in two species of Sphagnum. But given the wholly articial
and small-scale nature of the experiment, it seems difcult to
conclude as the authors do that High N deposition levels do
indeed appear to be responsible for the observed rapid veg-
etation changes in (actual) ombrotrophic bogs.This might
even be true, but guidance must be given by the authors of how
to conform the measures taken in articial environments and
apply them to actual environments, all while accounting for
the relevant uncertainties. However, this was not done.
Gunnarsson and Rydin
A fourth representative paper is
that describes an experiment
in which supplementary N from 0 to 10 g m
4Dose-Response: An International Journal
applied in Sphagnum hummocks and lawn communities on 20
× 20 cm plots. Ambient N ranged from 0.42 to 0.72 g m
. This was a rare paper that also included some indi-
cation of uncertainty of these background gures, for example,
standard deviation of ± 0.12 g m
for the ambient
The signal of the experiment was mixed: Sphagnum
showed an increased growth in length with the intermediate N
treatment, but in the second and third seasons the control
treatment had the highest growth in length.This was dis-
covered by the use of an ad hoc formula as a function of
length, density, and area in which samples were taken. That is,
the result of the output of this formula became the key ob-
servation of study.
The authors say that up to a point further N addition
reduces growth,but this could also be because of P or K
The authors also hazard a guess at a critical load, saying it is
, because this in their experiment matched the
optimalgrowth of Sphagnum. But this was not optimal
always and not everywhere. Growth at 5 g m
higher for both locations studied than 1 g m
, which
was similar to the growth at 10 g m
in these small
Shoot formation rates were maximized at 0 added g m
or three added g m
Again, all of these measures may be important, either taken
singularly, or taken together in some formulaic way. But that
they change under varying N is, by itself, not important. The
amount of change has to be demonstrated to be important by
other means or arguments.
Breeuwer et al.
The authors in
conducted a greenhouse experiment, exposing
different Sphagnum species collected at northern and southern
Swedish locations to either ambient N or ambient N plus 4 g
extra N. Temperature was also varied. The results
were mixed. Biomass production was calculated by a formula
with various plant measures as input. These formulas, which
many authors use, are not uniform, and have many variations.
This is not to argue any are right or wrong, as any of them
might be right for a particular decision. But results from
disparate formulas cannot be added together and directly
compared without consideration of the uncertainties involved.
Northern species (S. fuscum, S. balticum) height increment
and production changed with the changing temperature more
than did southern species (S. magellanicum and S. cuspida-
tum). The southern species height and production changed
more with the changing N. Some of the experimental con-
tainers suffered from severe fungal infection.
The ambient levels of atmospheric N the authors noted
from other publications, and were from 0.3 to 0.6 g m
. The ambient levels of N in the greenhouse was not
noted. Once again, the result is that change was noted with
varying levels of N, but that what constituted harm, or how to
extrapolate the small-scaled studies, was not laid out.
Limpens et al
The authors in
also ran an experiment in controlled glass-
house conditions, examining the growth of Betula pubescens
and Molinia caerulea in mixed Sphagnum samples gathered in
the Netherlands and given 40 kg ha
. They also
examined growth of seedlings or sprouts of the same species,
subject to infusions of 0, 40, or 80 kg ha
. There was
no apparent ambient level of N in the glasshouse atmosphere,
or none was listed. Plant production, as in other experiments,
was a formula based on various plant measurements. The same
comments to that formula apply as above.
As for results, the differences in shoot mass by N, for
instance, depended on the species (six were assessed). The
signals were mixed. For some species, shoot mass was highest
at 40 kg ha
, and others at 80 kg ha
, and one
(Molina) at 0 kg ha
. Presumably there must,
therefore, have been ambient glasshouse N. N in the interstitial
water changed through time in a non-linear manner.
Because most of the largest changes were in Molina, most
of the discussion revolved around it and not the other ve
species. This is not inappropriate if this is the key species in
some decision process. But it is also important to note that the
results differed widely by species. And, as remarked above, no
indication of how to extrapolate the experiment, which was in
an articial setting, to large scales was given.
Wiedermann et al
The work by
is the most important paper in our review of this
section because it attempted to answer how the small-scale
experiments might scale up to regional or country levels.
They divided Sweden into four unequally sized regions
with (from north to south) increasing levels of ambient at-
mospheric N, from which three mires were picked with a total
wet plus dry deposition of N at three, 11, and 16 kg ha
, with no uncertainty or variation given in these
numbers, though the authors admit the difculty of measuring
dry deposition.
A small-scale controlled eld experiment was also con-
ducted using 2 × 2 m plots from which 0.5 × 0.5 m samples
were taken. The experiments added-N at 0 (control), 2, 15, and
30 kg ha
. S was separately varied, though this is not
of direct interest to us. The ambient N was 2 kg ha
with no uncertainty given.
In the eld experiments, vascular plant coverage increased
on average in the samples from about 20% coverage at 2 kg
added-N, to about twice that at 30 kg ha
added-N. Total Sphagnum correspondingly decreased from
just under 100% to just over 20%. These obviously do not sum
to 100%, indicating these are relative area measures, and not
total plant cover.
Briggs and Hanekamp 5
Observations taken in situ at three Swedish locations were
also different with respect to coverages at the sites sampled. At
the northern site samples with 2 kg ha
(again, no
plus-or-minus to these were given), total Sphagnum was again
about 100%, dropping to about 70% at site samples with 12 kg
. Vascular plant cover rose from about 10% to just
under 40%.
The authors called the patterns of changing cover in both
the experiment and the in situ samples analogous patterns,
which they are. The directions of change are certainly sug-
gestive, but obviously many causal items of importance to
growth differed between the articial experiments and actual
samples. The patterns could just as equally well be described
by land use changes as by N differences. We do not claim this
is the case, only note that it could be so and that the exper-
iments did not rule this out.
Discussion of Bobbink and Hettelingh
From all these, and several other papers, Bobbink and Het-
telingh derived what they classed as NCL. The following, in
Table 1, is an adaptation of their Table 1 to indicate NCL for
EUNIS code D ecosystems and their reliability.
The Reliability codes are (we are quoting) ## reliable, #
quite reliable, (#) expert judgment. The ranges are not to be
taken as indications of uncertainty, but are used for different
subkinds of ecosystems, as indicated in their text. For ex-
ample, for raised and blanket bogs, the authors say use to-
wards high end of range at phosphorus limitation, and towards
lower end if phosphorus is not limitingor use towards high
end of range with high precipitation and towards low end of
range with low precipitation.These judgments rely, they say,
on several of the papers reviewed above (and of course others).
There is no other uncertainty in the numbers except for their
judgments about the reliability. This is susceptible to several
criticisms, which we lay out next.
(1) No clear indication on what critical means. The of-
cial denitions of what critical loads is are to some
extent clear with respect to political goals, but not
clear with respect to repeatable or consistent measures
on plant measurements or chemistry, for instance. We
can see from Table 1 the changes vary widely. This
isnt a problem if it is these, and only these, measures
upon which crucial decisions will be made. Other-
wise, the exact physical or biological state that denes
critical should be known and agreed upon before a
critical load in any substance can be discovered. Is it
growth rates of particular species beyond or below a
certain point? Is it the amount of accumulated detri-
tus? Of all plants? Only some? Is it a specic soil level
of N? A ratio of N to K or P in dried plant matter? All
species? Only some? It is a mix of species in which a
favored species is too high or too low?
There does not have to be one singular denition, because
there is not just one decision, or cost or benet, in relation to N
(or to anything biologically important chemical or biological
Whatever critical load is, it cannot change from one thing to
another; change in measure, that is. Change in itself is neither
good nor bad. That the different studies that showed, for
instance, how the length of a particular Sphagnum species is
changed on average in some way is not, by itself, of interest. It
must be specied why some level of growth, if only growth is
considered, is good or bad in some decisional manner.
The change has to be important in some named, clear, and
measurable way. Naturally, this denition of critical is ex-
pected to change in different regions and times of years, and it
might even vary by circumstances.
(2) Statistical confusion. Statistics can give meaningful
information about a critical load once it is dened,
especially in quantifying its variability, seasonality
and so forth. It should assist in quantifying uncertainty
in critical load exceedance measures, as in Nitrogen
Critical Load Modeling above. But one has to be
cautious in using it to dene critical loads itself.
There may be some confusion with statistical signi-
cance.In many of the experiments, biological measures were
compared between conditions with added-N (at many levels)
and a control, with N usually at ambient levels. If the dif-
ference between control and the other levels was statistically
signicant,that is a null hypothesis signicance test evinced
a wee P-value. This, by itself, is evidence of very little. Even
Table 1. Adapted from.
The Reliability codes are ## reliable, # quite reliable, (#) expert judgment. The ranges are not to be taken as
indication of uncertainty, but should be used for different subkinds of ecosystems, as indicated in their text.
Ecosystem Type kg ha
Reliability Indication of Exceedance
Raised and blanket bogs 510
Increase vascular plants, decrease bryophytes, altered growth and species
composition of mosses, increased N in peat and peat water
Poor fens 1015
Increase sedges and vascular plants, negative effects on peat mosses
Rich fens 1530
Increase tall graminoids, decrease diversity
Montane rich fens 1525
Increase vascular plants, decrease bryophytes
6Dose-Response: An International Journal
tiny differences, which would make no change in any possible
decision, can be signicant.AP-value is only that chance
that, if no differences existed, a test statistic would exceed
some level in new experiments (it actually means even less
than this, but this is close enough, see).
In any case, it is clear that statistical signicanceby itself
cannot be used to dene what is or is not critical. Critical has to
be dened outside of any statistical test, and the denition
must rely on the biology, chemistry, and even politics which
underlie the N system. Once criticalhas been objectively
dened, statistics can, of course, be used predictively and in
other ways in concert with observations. But testing can never
be the basis of the denition.
(3) Lack of realistic studies. Most of the papers relied on
very small areas in which precise measures were
taken, with the results of these implicitly extrapolated
to entire regions, or even whole countries. Other
experiments were wholly articial, inside glass houses
for instances. Its not that these experiments cannot
provide useful information for designing large-scale
measurements or experiments, but it cannot be argued
with force that what happens in a controlled 10 × 10 cm
plot is an error-free proxy for a region or country.
Even if what happens on these small plots can with con-
dence be extrapolated to large areas, that extrapolation comes
at a cost in certainty. It cannot be maintained that, say, the
exact proportion of species composition found on a 10 × 10 cm
plot with a rigorously controlled addition of N (with the other
components less well specied, or even unspecied) will be
duplicated at a regional or country level.
There are certain mathematical techniques that can provide
for the uncertainty due to extrapolations, but we would rec-
ommend against using them, because they themselves con-
stitute a model, and would be just as untested as using the
naive extrapolations. Again, large-scale experimentation is
called for.
(4) Full accounting of N. Nitrogen in soil less often
mentioned in most of the experiments, though there
are exceptions, for example,
Again, regional, local,
seasonal and other time-varying changes are never
accounted for. Sources are speculated upon, but few
conrmatory measures are taken. This, as we suggest
below, calls for greater experimentation and
(5) Expert judgment unscrutinized In,
critical loads for
various environments are given. For instance, for poor
fens, the critical load is estimated to be 1020 kg ha
. The measures relied up in these estimates are
said to be Increase sedges and vascular plants,
negative effects on peat mosses.The estimate is said
to be quite reliablebecause it depends on the sta-
tistical hypothesis testing, criticized above.
But in other instances, such as for rich fens, which have a
critical loads of 1535 kg ha
, and which relies on
increase tall graminoids, decrease diversity,the levels are
said to be made based on expert judgment. These judgments
also rely upon certain small-scale experiments, but which do
not have the same number of signicance testing.
rely on expert judgementoften in recom-
mending critical loads. There is an appearance of variability
noted in the judgments, but this is only apparent. The range is
only given to further distinguish sub-types of environments.
For instance, high latitude or nitrogen-limited systems
should use the low end of 1535 kg ha
. What exactly
is a nitrogen-limited system is not full specied.
We want to be clear. We do not wish to give the impression
these expert-guided limits or denitions are incorrect per se.
They may in fact be exactly correct. What we want to impart is
that, given the evidence and objections made thus far, the
certainty in these judgments is too high, and their bases too
ambiguous. There needs to be a way to verify their accuracy
(discussed), especially if and when expensive decisions will
be made relying on them.
More experimentation and observations are thus needed.
Toward a Better Understanding
of Uncertainty
The obvious intent of the studies discussed above is to infer
causal effects of N. This can be done, with the obvious
simplifying assumptions, for the small-plot experiments,
where there is some form of control, at least with respect to
the causative agent namely N. Extrapolating what was
learned from the small plots to vast regions is another thing at
together, of course, especially when trying to calculate the
Whats really needed, as we discuss below, are large-scale
experiments on the same space and time scales that are im-
portant to decision makers. A step in that direction has been
provided by
Both papers are similar, with
farther in an attempted correlative quantication of N and
certain biological measurements. The criticism we make is
that cause is inferred by these quantications, but should not
Its worth going through the report
in some detail, be-
cause it sets the worthy goal of discovering large-scale so-
called dose-response functions of atmospheric nitrogen and
certain plant observables, such as species richeness.
This makes a good rst start at such an effort, especially
considering much previous work, as we saw above, was on
small-scale (20 × 20 cm) highly controlled plots, which were
then extrapolated to regions, and even entire countries.
However, there are some areas of potential confusion
which are addressed here in the spirit of constructive debate as
much more work is needed to understand the science of N
related to dening NCL as decisionals requiring nationwide
Briggs and Hanekamp 7
Inappropriate Causal Inferences
The way both
argue is the following. They observe
various biological measures (such as total species richness)
at several points in several different ecosystems. They also
measure the corresponding N at those locations. They then
build models, by ecosystem, that quantify things like species
richness as a function of observed N.
They nd, roughly, that species richness decreases with
increasing N. Unfortunately, the temptation is to suppose that
Niscausing this decrease. This need not be so; indeed, it is
likely not so in certain situations, which we discuss below.
Nitrogen is implied as causing lack of species richness;
however, this might be an artefact of the way the data are
analyzed. The data taken from the UK (from)
is a good
example of this. This shows for the ecosystem of dry dune
grasslands, in total and separately for herbs and mosses, ni-
trogen deposition by total species richness.
We can here ignore the precise denition of total species
richeness.The suggestion that with and because of increasing
N deposition total species richenessdecreases is there in this
gure and in the text. But it is curious. The highest total
species richenessfor both herbs and mosses is highest at, it
appears, 510 kg N ha
. The only areas shown in
Figure 1 of Ref.
, which shows a map of N deposition of the
UK, where these levels of N are found are in far northern
Scotland, and a thin band of coast in western Northern Ireland.
These coastlines are, of course, where dunes are found, so it
is no surprise diversity of dune-based vegetation would be
highest nearest the coast, and tapering off as the dunes
transitions to other ecosystems.
Now the areas with least total species richenessis for
deposition rates of 1517 kg N ha
. These small areas
border the coasts, reaching inland only a small way, such as
middle northern Scotland, and on the inner western coast of
Northern Ireland, a few scattered areas in northern Yorkshire,
and a few very small areas along the coast of England. This is
judging only by the picture: there may be other areas, but the
point we make will be clear.
Obviously, as one moves inland, there will be fewer plant
species that are associated with dunes, since dunes are found on
coasts. It is therefore not necessarily N per se which is causing
the reduction of diversity, but the lack of inland dunes causing
declining dune-type vegetation. Certainly this alternate ex-
planation is plausible, if not a better explanation.
Other gures repeat this gradient studiesexercise for things
like moss species and lichens in heather-dominated moors
(Figure 5; p. 20)
and total vegetation and lichen in calcareous
grasslands in the UK (Figure 7; p. 22).
In some of these plots, species diversity does not appear to be
related to N because the standard deviation intervals overlap for
most level of N deposition. In others it might, as in Figure 3.
However, as Figure 1 of Ref.
illustrates, the same difculty in
assigning cause to N, and not to geography and land use, is found.
The conclusion that the een negative correlatie gevonden
tussen soortenrijkdom en/of samenstelling van de vegetatie en
stikstofdepositie(or a negative correlation was found be-
tween species richness and/or composition of vegetation and
nitrogen deposition) is not wrong, but it does remind us that
correlation is not causation, and should not be mistaken for it.
To turn this correlation into a causation, or at least turn it in
that direction, more controlled data is needed. For example,
nitrogen deposition rates at xed locations in which the de-
position rate is measured along species concentration in time.
If the one vary together in time, then a case can be made for
causation to be present. As it is, the signals so far seen could
just be natural variation.
Inappropriate Dose-Response Inferences
The concept of dose-responseis by denition causative. The
data itself, mentioned in the previous subsection, is only correl-
ative. Therefore, calling models of in situ observed N and things
like species richness, relies on the implication that correlation is
causation, that decreasing diversity is caused by increasing N.
Beyond all that, the so-called response curvesthemselves
(p. 106) are of interest. See Figure 2.
The blue dots are the
means of various biological measurements, with the standard
deviations given, and the N (x-axis) divides the depositions
into buckets in kg ha
, with no uncertainty given. The
dose-response curves are in black.
Instead of formally measuring curve t (there are many
methods), experts rated how well the curves represented the
data (presumably by eye). These judgments are indicated in
the colored dots at the top of the plots. Green is good, yellow
so-so, red bad.
Figure 2. A so-called dose-response curve from.
The x-axis is in
kg N ha
. The y-axis is species richness, the model is in
black, and the observations are in blue. The expert judgments are in
green and yellow at the top. The correlation ρmeasures agreement
on model t across experts.
8Dose-Response: An International Journal
The authors did not appear to use any formal method of
verication (the technical term for model diagnostics) because
they had created the curves by maximizing a correlation
coefcient. Perhaps they assumed formal methods of veri-
cation would thus be biased. This is somewhat, but not entirely
true. Calibration, for instance, could still be checked. As could
skill (measuring model performance against more simplistic,
or even a constant dose-response model).
In any case, as is pictured here, the model curves extend far
beyond the observed data. What is curious is that there is no
way to say this is a good t, or a bad t. In fact, it is no t at all.
There are only three data points. And here the dose-response
curve only comes close to 1 of these points. The curve is
estimated for N levels out past 20, up to 60, but the data only
goes to 15. But, even considering these criticisms, the curve
was judged Good by expert opinion.
Looking closer, the curve is off by 5 units at the third and
nal data point, which does not sound like a lot until its
considered the data itself only spans 10 units. The curve is thus
off by half here. The model cannot be considered good, but it
might not appear that poor because the curve extends far
beyond the observed N, stretching the curve.
The many small-scale studies done to identify critical loads of
nitrogen are either over-certain, inconsistent, or not conclu-
sive. Many are conducted on plotsaround the size of 20 ×
20 cm, inside glass houses, with the results extrapolated and
applied to large-scale regions, and even countries.
This obviously leads to great over-certainties. They are
made worse because most, or even all, of these studies
summarize ambient atmospheric nitrogen with one number,
regardless of the size of the region of country. No account of
variation, seasonality, or nitrogen sources is given.
The most difcult aspect is that these studies use inconsistent
denitions of what critical loadsare, dening them by crude
statistical tests of the small-scale experiments. In effect, any
change in these inconsistent outcomes over a controlis taken
to be critical,regardless whether the measures used (complex
formulas of plant size, for example) would be interesting,
actionable, or indeed cause any harm in actual locales.
There are two kinds of experiments that can and must be
done to rm up all these uncertainties: (1) observational, to
inform baselines and ranges, and (2) planned, to test causal
ideas learned in the rst step.
Nitrogen monitoring stations should be set up over a large area
to measure wet and dry atmospheric deposition, and soil N
content at several layers. The choice of location has to be for
both adequate coverage and to inform decisions about N that
must be made, as discussed shortly.
Those items thought or known to be associated with N must
also be monitored. Obviously, N by itself is not usually of
interest, but those things said to be causally affected by N.
Monitoring should be as nely grained as possible (large
number of stations), with measurements taken (automatically)
as frequently as possible. We need a clear indication of N ux,
its variability, seasonality, and other characteristics.
The locations should be near or at those areas that are
deemed crucial in some decisional sense, like agricultural
elds, to measure, for example, crop yields; fens or bogs, to
measure, for example, plant coverage or plant chemical
content; lakes and ponds, to measure, for example, weed
production. It must always be recalled that there are always
costs and benets from any policy. It should not only be costs
driving decisions. The benets of N should not be ignored, but
usually are in efforts to paint N as a pollutant only.
Also, a change in any measurable itemfrom one location
to the next, or one time to another, or because, say, farming has
commencedis not itself harmful or benecial. Deciding
harm or benet is something that is aboveor independent of
the measurements taken.
Certain atmospheric and soil variables should also be taken
along with N. Temperature, precipitation and wind at a mini-
mum, but also solar irradiance. Soil chemistry beyond N, such as
potassium and phosphorus, and soil moisture can also be gauged.
Spatial time-series statistics techniques, such as Kriging,
can be used to estimate N in those areas and times in which it
was not monitored. These techniques can also be used to
quantify uncertainty in the changes of biological measures
(such as crop yield or Sphagnum coverage) with changes in N.
Planned Experiments
Standard techniques for medium-to large-scale eld experi-
ments can be done, using blocking to add N to co-located elds
while keeping some as a control. The amounts added would be
by wet deposition, that is, adding to water and spraying.
This would be useful for gauging distance and nearness
effects of N to plots adjacent to experimental elds. The
differences, if any, between biologically relevant measures in
adjacent plots can also be ascertained. That is, the differences,
if any, between those plots adjacent to where N was added, and
those plots adjacent to control plots.
In this way the geographic extent of adding known amounts
of N can be quantied.
Thanks to Geesje Rotgers for commenting on an earlier draft of this
Declaration of Conicting Interests
The author(s) declared no potential conicts of interest with respect to
the research, authorship, and/or publication of this article.
Briggs and Hanekamp 9
The author(s) received no nancial support for the research, au-
thorship, and/or publication of this article.
Jaap Hanekamp
1. Van Dobben H, Bobbink R, Bal D, Van Hinsberg A. Overzicht
Van Kritische Depositiewaarden Voor Stikstof, Toegepast Op
Habitattypen En Leefgebieden Van Natura. Technical Report
Alterra-rapport 2397, Wageningen. Wageningen, Netherlands.
Wageningen Municipality; 2012.
2. Vries WD, Hettelingh J, Posch M. Critical Loads and Dynamic
Risk Assessments Nitrogen, Acidity and Metals in Terrestrial
and Aquatic Ecosystems. Environmental Pollution, Volume 25.
Dordrecht, Netherlands: Springer; 2015.
3. Nilsson J, Grennfelt P. Critical Loads for Sulphur and Nitrogen.
Report from a Workshop, Skokloster, Sweden, 19-24 mar 1988.
Technical report. Copenhagen, Denmark: Nordisk Ministerraad;
4. Heer MD, Roozen F, Maas R. The integrated approach to ni-
trogen in the netherlands: A preliminary review from a societal,
scientic, juridical and practical perspective. J Nat Conserv.
5. Banin L, Bealey B, Smith R, Sutton M, Campbell C, Dise N.
Quantifying Uncertainty in Critical Loads. Technical report,
CEH Report to SEPA. Lancaster, UK: Centre for Ecology &
Hydrology; 2014.
6. Briggs WM, and Hanekamp J. Outlining a new method to
quantify uncertainty in nitrogen critical loads. J. Philosophy
7. Aerts R, Wallen B, Malmer N. Growth-limiting nutrients in
sphagnum-dominated bogs subject to low and high atmospheric
nitrogen supply. J Ecol. 1992;80:11-39.
8. Wiedermann MM, Gunnarsson U, Ericson L, Nordin A. Eco-
physiological adjustment of two sphagnum species in response to
anthropogenic nitrogen deposition. New Phytol. 2009;181:208-217.
9. Breeuwer A, Heijmans MMPD, Gleichmanand M, Robroek
BJM, Berendse F. Response of sphagnum species mixtures to
increased temperature and nitrogen availability. Plant Ecol.
10. Tomassen HBM, Smolders AJP, Lamers LPM, Roelofs JGM.
Stimulated growth of betula pubescens and molinia caerulea on
ombrotrophic bogs: role of high levels of atmospheric nitrogen
deposition. J Ecol. 2003;91:357-370.
11. Bobbink R, Hettelingh J-P. Review and revision of empirical
critical loads and dose-response relationships. In: Proceedings of
an Expert Workshop, Noordwijkerhout, Netherlands, 23-25
June 2010. Bilthoven, Netherlands: Coordination Centre for
Effects, National Institute for Public Health and the Environ-
ment (RIVM); 2011.
12. Bragazza L, Tahvanainen T, Kutnar L, et al. Nutritional con-
straints in ombrotrophic sphagnum plants under increasing at-
mospheric nitrogen deposition in europe. New Phytol. 2004;163:
13. Berendse F, Van Breemen N, Rydin H, et al. Raised atmospheric
CO2 levels and increased N deposition cause shifts in plant
species composition and production in Sphagnum bogs. Global
Change Biol. 2001;7:591-598.
14. Gunnarsson U, RydinH. Nitrogen fertilization reduces sphagnum
production in bog communities. New Phytol. 2000;147:527-537.
15. Limpens J, Berendse F, Klees H. N deposition affects N
availability in interstitial water, growth of sphagnum and in-
vasion of vascular plants in bog vegetation. New Phytol. 2002;
16. Wiedermann MM, Gunnarsson U, MatsNilsson ANB, Ericson
L. Can small-scale experiments predict ecosystem responses?
An example from peatlands. Oikos. 2008;118:449-456.
17. Briggs WM. Uncertainty: The Soul of Probability, Modeling &
Statistics. New York, NY: Springer; 2016.
18. Berendse F, Lammerts EJ, Olff H. Soil organic matter accu-
mulation and its implications for nitrogen mineralization and
plant species composition during succession in coastal dune
slacks. Plant Ecol. 1998;137:71-78.
19. Field CD, Dise NB, Payne RJJ, Btitton AJ, et al. The role of
nitrogen deposition in widespread plant community change
across semi-natural habitats. Ecosystems 2014;17(5):864-877.
20. Wamelink G, Goedhart P, Roelofsen H, Bobbink R, Posch M,
van Dobben H. Relaties Tussen De Hoeveelheid Stik-
stofdepositie En De Kwaliteit Van Habitattypen. Technical
Report nr. 228471006. Wageningen, The Netherlands: Wage-
ningen Environmental Research; 2021.
10 Dose-Response: An International Journal
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
In north-western Europe, deposition of atmospheric nitrogen is one of the main obstacles to maintaining or restoring natural habitats to a favourable conservation status. The Integrated Approach to Nitrogen (PAS) of the Netherlands is a national plan that combines generic source measures to reduce nitrogen emission levels and ecological restoration measures in Natura 2000 areas, while creating room for economic development. The aim of the PAS is to ensure that conservation objectives can be achieved, while further economic development is facilitated around Natura 2000 areas within strict environmental limits. In this way, the PAS connects economy and ecology. This paper examines the PAS from a societal, scientific, juridical and practical perspective, based on a study of the literature, juridical cases and first experiences of the programme. Our review indicated that the PAS is a comprehensive approach to the nitrogen issue, aiming for a balance between the societal, scientific, juridical and practical perspective. However, the programme has not yet been in force long enough to observe actual results in the field.
Full-text available
In dit rapport wordt een overzicht gegeven van concrete (unieke) kritische depositiewaarden voor stikstof voor de habitattypen en de (stikstofgevoelige) overige leefgebieden van soorten die in Natura 2000-gebieden worden beschermd. Hiertoe zijn de door de UNECE in 2010 vastgestelde kritische depositiewaarden voor stikstof nader gepreciseerd en (voor zover nodig) aangevuld, waarbij gebruik is gemaakt van modeluitkomsten en deskundigenoordeel. Het rapport is een actualisering en uitbreiding van een eerdere versie (Van Dobben en Van Hinsberg, 2008).
Full-text available
Experimental studies have shown that deposition of reactive nitrogen is an important driver of plant community change, however, most of these experiments are of short duration with unrealistic treatments, and conducted in regions with elevated ambient deposition. Studies of spatial gradients of pollution can complement experimental data and indicate whether the potential impacts demonstrated by experiments are actually occurring in the 'real world'. However, targeted surveys exist for only a very few habitats and are not readily comparable. In a coordinated campaign, we determined the species richness and plant community composition of five widespread, semi-natural habitats across Great Britain in sites stratified along gradients of climate and pollution, and related these ecological parameters to major drivers of biodiversity, including climate, pollution deposition, and local edaphic factors. In every habitat, we found reduced species richness and changed species composition associated with higher nitrogen deposition, with remarkable consistency in relative species loss across ecosystem types. Whereas the diversity of mosses, lichens, forbs, and graminoids declines with N deposition in different habitats, the cover of graminoids generally increases. Considered alongside previous experimental studies and survey work, our results provide a compelling argument that nitrogen deposition is a widespread and pervasive threat to terrestrial ecosystems.
Full-text available
We studied the effects of N deposition on shrubmoss competition and the establishment and growth of invasive Betula pubescens and Molinia caerulea in intact bog vegetation removed from a site subject to 40 kg N ha1 yr1. Mesocosms with and without introduced Betula seedlings and Molinia sprouts were kept under a roof and received an equivalent of 0, 40 and 80 kg N ha1 yr1 for two growing seasons. N concentration in both interstitial water and Sphagnum decreased when N input ceased and increased when N input was doubled. Molinia biomass was positively related to the inorganic N concentration in the interstitial water. Adding N increased production of Molinia and prolonged survival of Betula seedlings in the first year. Sphagnum height increment showed a hump-shaped relationship with light interception by vascular plants. N deposition encouraged vascular plants to grow by enhancing N availability in the rhizosphere. Water table level and the availability of P were found to be important in explaining species-specific responses to N deposition. The underlying mechanisms and the reversibility of N effects are discussed
This book presents a philosophical approach to probability and probabilistic thinking, considering the underpinnings of probabilistic reasoning and modeling, which effectively underlie everything in data science. The ultimate goal is to call into question many standard tenets and lay the philosophical and probabilistic groundwork and infrastructure for statistical modeling. It is the first book devoted to the philosophy of data aimed at working scientists and calls for a new consideration in the practice of probability and statistics to eliminate what has been referred to as the "Cult of Statistical Significance". The book explains the philosophy of these ideas and not the mathematics, though there are a handful of mathematical examples. The topics are logically laid out, starting with basic philosophy as related to probability, statistics, and science, and stepping through the key probabilistic ideas and concepts, and ending with statistical models. Its jargon-free approach asserts that standard methods, such as out-of-the-box regression, cannot help in discovering cause. This new way of looking at uncertainty ties together disparate fields — probability, physics, biology, the “soft” sciences, computer science — because each aims at discovering cause (of effects). It broadens the understanding beyond frequentist and Bayesian methods to propose a Third Way of modeling. • Presents a complete argument showing why probability should be treated as a part of logic • Broadens understanding beyond frequentist and Bayesian methods, proposing a Third Way of modeling • Proposes that p-values should die, and along with them, hypothesis testing William M. Briggs, PhD, is Adjunct Professor of Statistics at Cornell University. Having earned both his PhD in Statistics and MSc in Atmospheric Physics from Cornell University, he served as the editor of the American Meteorological Society journal and has published over 60 papers. He studies the philosophy of science, the use and misuses of uncertainty - from truth to modeling. Early in life, he began his career as a cryptologist for the Air Force, then slipped into weather and climate forecasting, and later matured into an epistemologist. Currently, he has a popular, long-running blog on the subjects written about here, with about 70,000 - 90,000 monthly readers.
1. The effects of increased nitrogen or phosphorus supply on the productivity of Sphagnum-dominated ombrotrophic bogs in northern and southern Sweden were studied. Atmospheric nitrogen deposition in southern Sweden (high-N site) exceeds that in northern Sweden (low-N site) by about tenfold. 2. Vertical height growth of the Sphagnum carpet was measured by the cranked-wire method. Length growth of individuals was measured by autoradiography after labelling with 14CO2. The results of both methods were significantly correlated, but the cranked-wire data were systematically lower. 3. Productivity of Sphagnum at the low-N site increased almost fourfold after additional nitrogen supply (4 g N m-2 year-1), but no increase was found after additional phosphorus supply (0.4 g P m-2 year-1). At the high-N site, phosphorus supply caused an almost threefold productivity increase, but nitrogen supply did not result in any productivity increase. Thus, in an area with a high atmospheric nitrogen supply, plant productivity is P-limited instead of N-limited. 4. At an intermediate nitrogen supply (2 g N m-2 year-1) the recovery of the supplied nitrogen in the Sphagnum carpet was not different for both sites (60% and 69%, respectively). However, at a high nitrogen supply (4 g N m-2 year-1) nitrogen recovery at the low-N site significantly exceeded that at the high-N site (73% and 47%, respectively). At the low-N site, the supplied phosphorus was not recovered, but at the high-N site the phosphorus recovery was 85% (intermediate phosphorus supply) and 100% (high phosphorus supply), respectively. 5. It is suggested that a high atmospheric nitrogen supply may affect the carbon balance of ombrotrophic bogs, because productivity is under these circumstances not N-limited, but decomposition is probably increased by high loads of nitrogen. In the end, this may turn these C-accumulating systems into C-emitting systems.
A group of Nordic experts has tried to draw conclusions on critical loads for sulphur and nitrogen. The critical load is defined as “The highest load that will not cause chemical changes leading to long-term harmful effects on most sensitive ecological systems”. Most soils, shallow groundwaters and surface waters would probably not be significantly changed by a load of 10–20 keq H+·km2·yr−1 in areas with a low content of base cations in the deposition. The total deposition of hydrogen ions in southwestern Scandinavia is in the order of 100 keq ·km−2·yr−1. The long-term critical load for nitrogen is in the range of 10–20 kg N·ha·1-yr−1 in most forest ecosystems. In high productive sites it might be as high as 20–45 kg N·ha− yr−1 in southern Sweden, and amounts to 30–40 kg·ha−1·yr−1 and even more over large areas in central Europe. The current deposition of sulphur and nitrogen must be substantially reduced to keep the long-term changes in sensitive ecosystems within acceptable limits.
1 In order to test whether the observed invasion of ombrotrophic bogs in the Netherlands by Molinia caerulea and Betula pubescens is the result of long-term high nitrogen (N) loads, we conducted a 3-year fertilization experiment with Sphagnum fallax turfs. Six different N treatments were applied ranging from 0 (control) to 4 g N m−2 year−1. 2 During the experimental period, ammonium concentrations in the peat moisture remained very low due to high uptake rates by Sphagnum. Tissue N concentrations in S. fallax showed a linear response to the experimental N addition. Excess N was accumulated as N-rich free amino acids such as arginine, asparagine and glutamine, especially at N addition rates of 0.25 g m−2 year−1 or higher, indicating N-saturation. 3 Despite the high tissue N : P ratio (above 35), above-ground biomass production by Molinia was still stimulated at N addition rates of 4 g m−2 year−1, and foliar nutrient concentrations were unaffected compared to the control. In contrast to Molinia, Betula was unable to increase its above-ground biomass. Foliar N concentrations in Betula were significantly higher at N addition rates of 4 g m−2 year−1 and excess N was stored in foliar arginine, making up 27% of the total N concentration. Evapotranspiration was increased at higher N addition rates due to stimulated total above-ground biomass production of the vegetation. 4 N addition at the actual Dutch deposition rate of 4 g m−2 year−1 stimulated the growth of Molinia in this experiment, providing evidence that the observed dominance of Molinia on ombrotrophic bogs in the Netherlands is caused by high N deposition levels. Based on the observed changes in biomass production and tissue nutrient concentrations, we assume that a long-term deposition of 0.5 g N m−2 year−1, or higher, leads to undesirable changes in species composition and increased risk of desiccation.
The effects of increased nitrogen influx on Sphagnum growth and on interspecific competition between Sphagnum species were studied in a 3-yr experiment in mires situated in two areas with different rates of airborne N deposition. Sphagnum growth was recorded after various supplementary N influxes (0, 1, 3, 5 and 10 g m −2 yr−1) in hummocks and lawn communities. Sphagnum biomass production decreased with increasing N influx in both areas. After the first season at the low-deposition site, Sphagnum showed an increased growth in length with the intermediate N treatment, but in the second and third seasons the control treatment had the highest growth in length. Capitulum dry mass increased with increasing N influx. Sphagnum N concentration and N/P quotient were higher at the high- than at the low-deposition site. The low quotient at the low-deposition site, together with the initial growth increase with intermediate N supplements, indicates that growth was N-limited at this site, but our lowest N supplement was sufficient to reduce growth. The N treatments had no effect on interspecific competition between the Sphagnum species. This indicates that the species have similar responses to N. The species studied all occur naturally on ombrotrophic, N-poor sites and show low tolerances to increased N influx. Reduced Sphagnum production may affect the carbon balance, changing the mires from C sinks to sources.