Journal of Environmental Management 206 (2018) 650e663

Contents lists available at ScienceDirect

Journal of Environmental Management

journal homepage: www.elsevier.com/locate/jenvman

Research article

Analysing the impacts of air quality policies on ecosystem services; a

case study for Telemark, Norway
L. Hein a, *, L. White b, A. Miles b, P. Roberts c
a
Environmental Systems Analysis Group, Wageningen University, P.O. Box 47, 6700 AA, Wageningen, The Netherlands
b
Aeris Europe, Strouds, Church Lane, Horsted Keynes, RH17 7AY, UK
c
Concawe, Boulevard du Souverain 165, B-1160, Brussels, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: There is an increasing interest in considering the effects of air pollution on ecosystem services supply in
Received 24 May 2017 order to enhance cost-benefit analyses of air pollution policies. This paper presents a generic, conceptual
Received in revised form approach that can be used to link atmospheric deposition of air pollutants to ecosystem services supply
1 September 2017
and societal benefits. The approach is applied in a case study in the Telemark county of Norway. First, we
Accepted 30 October 2017
Available online 10 November 2017
examine the potential effects of four European air quality policy scenarios on N deposition in the eco-
systems of this county. Second, we analyse the subsequent impacts on the supply of three ecosystem
services: carbon sequestration, timber production and biodiversity. Changes in the supply of the first two
Keywords:
Air pollution
services are analysed in both physical and monetary units, biodiversity effects are only analysed in
Ecosystem services physical terms. The scenarios derive from work conducted in the context of the European National
Nitrogen deposition Emissions Ceilings Directive. In the 2010 base case the benefits of carbon sequestration are estimated at
Norway 13 million euro per year and the value of timber harvesting at 2.9 million euro per year. Under the
examined policy scenarios aiming to reduce nitrogen emissions the societal benefits resulting from these
two ecosystem services in Telemark are found to be reduced; the scenarios have little effect on terrestrial
biodiversity. Such results cannot be scaled up, individual ecosystem services respond differently to
changes in air pollution depending upon type of pollutant, type of ecosystem, type of service, and the
magnitude of change. The paper further presents an analysis of the uncertainties that need to be
considered in linking air pollution and ecosystem services including those in deposition rates, ecosystem
responses, human responses and in the values of ecosystem services. Our conceptual approach is also
useful for larger scale analysis of air pollution effects on ecosystem services, for example at national or
potentially European scale.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction
In recent years, there has been an increasing consideration of
ecosystem services in environmental policy making (e.g. Daily et al.,
2009; Maes et al., 2012). Ecosystem services have been defined as
the contributions of ecosystems to human well-being and include
such diverse aspects as products obtained from ecosystems
(including crops from agricultural land), the regulation of biological
processes (e.g. pollination, climate regulation) and non-material
services including tourism and recreation (e.g. TEEB, 2010; UN
et al., 2014). Ecosystem services can be expressed in both physical
* Corresponding author.
E-mail address: lars.hein@wur.nl (L. Hein).
and monetary units and allow considering the effects of changes in
the environment in cost-benefit analysis (Daily et al., 2009).
Until recently, the general approach in providing policy advice
on the effects of reduced air pollution is to compare actual or future
deposition levels with critical loads (see e.g. EEA, 2015). Critical
loads are defined as the highest load that will not cause chemical
changes leading to long-term harmful effects on most sensitive
ecosystems (Nilsson, 1988; Hettelingh et al., 2007). However this
damage based approach does not convey any information on the
societal benefits of reducing air pollution loading (De Smet et al.,
2007).
Hence, increasingly, ecosystem services are connected with air
pollution. For example, the European Environment Agency analyses
air pollution effects on ecosystems (EEA, 2015) and provides an
estimate of the effects of ozone (O3) deposition on timber

https://doi.org/10.1016/j.jenvman.2017.10.073
0301-4797/© 2017 Elsevier Ltd. All rights reserved.
 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663
production. Smart et al. (2011) applied an ecosystem services
approach to analyse the benefits from air pollution control in the
UK at the local scale. Jones et al. (2014) present a national scale
analysis of the impacts of air pollution on ecosystem services in the
UK, including the impacts of air pollution on wheat production and
biodiversity. These studies show the importance of considering the
multiple services provided by ecosystems as well as the spatial
variation in deposition and ecosystem responses. Both vary
strongly across the landscape and so do changes in ecosystem
services supply as a function of reductions or increases in deposi-
tion (e.g. Jones et al., 2014). Although the number of studies in this
field is growing there are still few that systematically link air
pollution to multiple ecosystem services in a spatially explicit
manner.
The objective of this paper is to examine how ecosystem ser-
vices can be assessed in support of air quality policy formulation,
using eutrophication in Telemark, Norway as a case study. We
propose a generic, spatially explicit approach, and apply it to
analyse the effects of atmospheric nitrogen deposition on carbon
sequestration, timber production and biodiversity conservation.
The approach combines an impact pathway approach to analysing
impacts of air pollution (e.g. Dominici et al., 2010; Thompson et al.,
2014) with an ecosystem services approach to assess impacts of air
pollution (Smart et al., 2011) e in a spatially explicit manner. In
addition, we analyse the methodological challenges and un-
certainties in linking air pollution and ecosystem services.
We select Telemark in view of the sensitivity of European boreal
forests and alpine ecosystems to eutrophication (Bobbink et al.,
1998; De Vries et al., 2014) and in view of the large amount of
data that are available on ecosystem services supply in Telemark,
both from the detailed statistics of Statistics Norway, and from
various papers (Schro € ter et al., 2014). We focus on nitrogen because
boreal forests tend to react strongly to nitrogen (De Vries et al.,
2014). We conduct our study for a specific county of 15,300 km2
because we want to test an approach involving high resolution
spatial modelling. Our study focusses on the EU, but is of wider
relevance given the occurrence of air pollution in other parts of the
world.
2. Methodology
2.1. Including ecosystem services in air quality assessments
Commonly considered pollutants in the context of European air
quality policy making are SOx, NOx, N2O, NH3, Particulate Matter
(PM), O3 and VOCs. These pollutants are emitted in different ways,
from near-surface sources such as exhaust pipes of cars or pipes of
household fireplaces to industrial smokestacks. The pollutants
follow different atmospheric dispersion pathways according to the
height of the source, chemical transformations in the atmosphere,
and the pollutant deposition velocity. For example NH3 from agri-
culture is generally deposited close to the emission source whereas
PM, NOx and SOx emitted from tall stacks travel larger distances
(Amann et al., 2011).
Analysing how these pollutants affect ecosystems and influence
ecosystem services supply requires consideration of the causal chain
linking emissions to ecosystem service impacts following an Impact
Pathway Approach (e.g. EPA, 2011). As a first element, this requires a
description of emissions followed by air pollution modelling to
calculate air concentrations and deposition. There are two deposi-
tion mechanisms. Wet deposition accounts for the effect of rainfall
depositing pollutants in the ecosystem. Dry deposition depends on
both the physical transfer of material to the surface and on uptake at
the surface. Dry deposition depends on the pollutant and on the
properties of the surface and so varies with ecosystem type.
651
The subsequent ecosystem impacts depend upon the mix of
pollutants the ecosystem is exposed to SOx and NOx are main
contributors to the acidification of ecosystems, whereas NOx and
NH3 contribute to eutrophication. Not all pollutants entering the
ecosystem are available for plant uptake, depending upon leaching
and absorption in the soil complex. The response of the ecosystem
to the pollutant is likely to be complex and can be characterised by
dose-response functions. Whereas these functions may be assumed
to be linear for specific pollution concentrations, non-linear
reponses will often occur across the range of exposure (Aber
et al., 1998; Smith, 1990).
Ecosystem change is usually a function of the combined effect of
different drivers, and can only be meaningfully captured with
multiple indicators for the state of the ecosystem, such as species
composition, crown cover and/or pH of the (ground) water. Critical
loads are an important concept in the assessment of ecosystem
effects of air pollution (Amann et al., 2011). Critical loads represent
the occurrence of thresholds in soils at which a rapid decrease in pH
can be expected (due to the occurrence of calcium and aluminium
buffers). At the point in time pollutant exposure levels start
exceeding critical loads rapid changes in species composition (e.g.
shifts in plant communities) and ecosystem functioning may occur,
with implications for biodiversity, the ecosystem's capacity to
supply services as well as ecosystem resilience, i.e. responses of the
ecosystem to future stressors.
Importantly, a change in ecosystem condition, for instance when
a critical load is exceeded, does not necessarily directly affect the
supply of ecosystem services. Rather, changes in ecosystem
composition and/or ecosystem functioning may affect the capacity
of the ecosystem to supply services. The capacity of an ecosystem to
supply services is defined as ‘the ability of an ecosystem to generate
a service under current ecosystem condition and uses, at the
highest yield or use level that does not negatively affect the future
supply of the same or other ecosystem services from that
ecosystem’ (Hein et al., 2016a,b). Capacity indicates the maximum
sustainable use level of the ecosystem subject to there being a
demand for the ecosystem services involved (Hein et al., 2016a,b).
The supply of ecosystem services depends upon the use of the
ecosystem by people. If there is a change in ecosystem capacity to
supply a specific service, people may respond by increasing or
reducing the amount of service used (for instance to bring use of
the service in line with sustainable use levels) e or they may not
respond and continue with the same use level. In case capacity is
reduced due to for instance air pollution such that present use
levels exceed the capacity, this will lead to ecosystem degradation
(Villamagna et al., 2013).
Generally, three types of ecosystem services are distinguished,
i.e. provisioning, regulating and cultural services (e.g. MA, 2005;
TEEB, 2010; Haines-Young and Potschin, 2012). Changes in
ecosystem condition will affect the capacity of the ecosystem to
generate the three types of services in different ways. The capacity
of ecosystems to supply provisioning services generally depends
upon the stocks and the annual increment of the species involved.
Changes in condition resulting from pollution may affect the stock
as well as the annual increment. Regulating services depend on
ecological processes that take place at specific temporal and spatial
scales, and often these are subject to naturally occurring fluctua-
tions in addition to human stressors such as air pollution. The ca-
pacity to generate regulating services may be affected by changes in
these processes. Cultural services include the non-material benefits
provided by ecosystems. Recreation and tourism are the two cul-
tural services that are most commonly included in ecosystem as-
sessments (TEEB, 2010) and the capacity to support them may be
affected by pollution leading to changes in the attractiveness of the
landscape.
652 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663
A generic approach that can be used to analyse the effects of air
pollution on ecosystem services is presented in Fig. 1. The specific
models and methodologies required to apply the approach need to
be developed as a function of types of pollutant included in the
analysis, study area, ecosystem services considered and data
availability. We limit our analysis to the monetary impacts of
eutrophication of terrestrial ecosystems in Telemark as a function
of atmospheric deposition, and the subsequent impacts on the
three selected ecosystem services.
2.2. Case study area
Telemark County is located in southern Norway. It covers around
15,300 km2 and has a population of about 170,000 people. The
county has a diverse landscape ranging from the sea to several
mountains exceeding 1500 m in elevation. The high-altitude alpine
areas have extensive snow cover in winter and have sparse vege-
tation mostly consisting of dwarf trees, shrubs, herbs and mosses.
The lower altitude coastal area has a more moderate climate, with
vegetation consisting of pine, spruce and birch forests. Annual
rainfall varies from 800 to 1000 mm depending upon altitude and
location. Agriculture is concentrated on alluvial deposits along the
rivers flowing through the county, and seasonal reindeer herding
and livestock grazing (mostly sheep) takes place in the uplands.
2.3. Analysis of the effects of nitrogen deposition in Telemark
2.3.1. Step 1. analyse the drivers for ecosystem change
Wet and dry depositions of nitrogen have been calculated using
source-receptor (S-R) functions. These S-R functions were gener-
ated through the EMEP/MSC-W chemical transport model. The
EMEP/MSC-W chemical transport models developed at the Mete-
orological Synthesizing Centre - West (MSC-W) are concerned with
the regional atmospheric dispersion and deposition of the acidi-
fying and eutrophying compounds Sulphur (S) and Nitrogen (N),
ground level ozone (O3) and particulate matter (PM2.5, PM10) .1 The
S-R function represents the change in concentration and deposition
in an area due to emission changes, relative to a baseline. It is based
1
For more information please visit: http://www.emep.int/mscw/mscw_models.
html.
 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663
Fig. 1. Framework for analysing the effects of air pollution on ecosystem services.
on many hundreds of atmospheric dispersion calculations deter-
mining the individual effect of perturbing emissions (NH3, NMVOC,
NOX, PM2.5, PMCO, SOX) from European countries. The calculations
are made for five years (2006e2010) and an average of the outputs
is taken in order to account for the effect of inter-annual meteo-
rological variation. For our study, modelling data were kindly pro-
vided by the Norwegian Meteorological Institute.
The emission scenarios we used are those used in the Clean Air
for Europe Program of the European Commission to examine options
for revising the National Emission Ceilings Directive (IIASA, 2014;
IIASA, 2016), see Table 1. The baseline scenario is grounded in
2010 and mirrors the emissions in time that would be expected with
economic development and legislation in place in 2010. Our paper
models this 2010 baseline scenario as well as two scenarios for 2020
and two for 2030, the emissions for which have been produced by
the IIASA GAINS model. For 2020 we analyse the Working Party on
the Environment (WPE) Current Legislation (CLE) scenario, and a
scenario representing full implementation of the revised Gothen-
burg protocol. The primary difference between these two scenarios
is the underlying PRIMES energy projection. PRIMES is a partial
equilibrium model of the energy systems within a country, devel-
oped for all countries in the EU (Capros et al., 2011). It is used to
make projections on energy prices, supply and demand, for all sec-
tors and technologies. The PRIMES 2009 projection utilized by the
Gothenburg protocol predicts a higher overall activity by 2020 than
the PRIMES 2013 projection that the other scenarios in this paper
use. For the year 2030, we again analyse the WPE CLE scenario as
well as the IIASA “optimal” scenario, which assumes implementa-
tion of all air pollution control measures deemed cost-effective in
the GAINS model to achieve an overall emissions reduction of 67%. A
full description of the scenarios can be found in IIASA (2014).
The emissions data were used to generate the mass of nitrogen
deposition as deposition averages per scenario for each of the 29
EMEP grids2 that cover Telemark (see Appendix 1). For the 2010
2
Each grid measures 0.5 longitude  0.25 latitude, this broadly equates to
28 km  28 km, almost twice the resolution of the standard 50 km  50 km EMEP
grid.
653
Table 1
Emissions scenarios used in this paper.
Scenario Energy Projection Years Modelled
WPE ‘TSAP 16’ CLE PRIMES 2013 2010, 2020, 2030
Gothenburg Protocol PRIMES 2009 2020
WPE ‘TSAP 16’ OPT (67% reduction) PRIMES 2013 2030
baseline and each of the four scenarios, deposition of oxidized and
reduced nitrogen are derived from the EMEP model. Deposition
rates are retrieved for each type of ecosystem.
2.3.2. Step 2. analyse ecosystem responses
Effects of nitrogen deposition on ecosystems have been studied
with modelling studies (e.g. Wamelink et al., 2009) and nitrogen
addition experiments (e.g. Hogberg et al., 2006). These studies
consider that part of the nitrogen entering the ecosystem will run-
off to surface water or leach to groundwater. Effects on ecosystems
are driven by the combined effect of the potential eutrophying and
acidifying effects of nitrogen deposition (De Vries, 2014). The
combined effects of eutrophication and acidification are complex
and strongly dependent upon the ecosystem context (climate, soil,
species composition, etc.).
We assume that there are no negative effects of acidification due
to nitrogen deposition on forest productivity in Telemark, based on
Hogberg et al. (2006), who states that current low levels (<10 kg N/
ha/year) of nitrogen deposition in northern Europe are unlikely to
lead to acidification affecting plant and ecosystem productivity. At
past and current ambient levels of N deposition in Northern Europe,
these ecosystems will remain strongly N limited for a long time and
there is no direct evidence that soil acidification at current rates in
Boreal forest leads to detrimental effect on tree growth (Binkley
and Ho €gberg, 1997).
The effects of eutrophication on ecosystems will generally be
the largest in ecosystems that are nitrogen constrained, i.e. where
nitrogen is the main limiting factor to plant growth, which includes
major parts of boreal ecosystems. A recent meta-analysis showed
that the responses of boreal forests to N input vary between 16 and
25 kg C kg N
1 (De Vries et al., 1994). This range is based upon an N
654 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663
retention (percentage of N deposited that is retained in the vege-
tation) of 25%. The analysis also shows that the ecosystem response
to N deposition in boreal forest, measured as the kg C kg N
1 ratio,
is typically some 50% higher compared to the response in temporal
forest, and a factor 3 higher compared to tropical forests. The latter
reflects that temperate and tropical forests include a larger amount
of nitrogen fixing species, and that tropical forests are more typi-
cally P rather than N limited. However, the relation between N
deposition and C accumulation in forest ecosystems is complex. For
example, at higher N deposition rates the effect per unit of N tends
to become smaller, and the effect of N deposition also varies with
the age of the forest stands, generally becoming lower for older
stands and older rotations in production forest (De Vries et al.,
1994). We come back to this in the Discussion section.
Subsequently, indicators have to be selected that can be linked
to the ecosystem's capacity to generate services. For our study, we
select Net Ecosystem Productivity (NEP), expressed in kg C/ha/year,
as an indicator for annual biomass growth in ecosystems. NEP
equals the gross primary productivity (the total carbon that pri-
mary producers accumulate using photosynthesis) minus the
autotrophic respiration (carbon respired by primary producers)
minus the heterotrophic respiration (carbon respired by other or-
ganisms, usually by detritus feeders that decompose dead material
in the soil). We modelled the relation between N deposition and
NEP, as well as subsequent steps of the spatial analysis, in ArcGIS
10.3.1.
We limit our analysis to effects on terrestrial ecosystems. We do
not further analyse the effect of the deposition of nitrogen on the
acidification of surface water in Telemark. Telemark has been one of
the Norwegian counties that suffered most from acid rain in the last
decades, with regular liming taking place in 339 lakes (Austnes,
2013). The main factor contributing to acidification was the
sulphur deposition leading to the formation of sulphuric acid.
Sulphur deposition rates have declined considerably since the
1980s because of successful air quality policies, and recent nitrogen
deposition levels in Telemark are too low to be a major contributor
to acidification (Skjelkvåle et al., 2001). We also do not consider the
effect of eutrophication on lakes, given that eutrophication of
Norwegian water courses is driven mostly by agriculture and that
phosphorus generally is the main limiting factor in these lakes
(Bechmann et al., 2005). We also do not consider the (positive)
effects of nitrogen deposition on cropland in Telemark, given that
the atmospheric deposition of nitrogen in cropland in this county is
generally small (on average <3 kg N/ha/year in the 2010 baseline)
compared to the amount of nitrogen fertilizer added by farmers.
2.3.3. Step 3 identify adaptation options and analyse effects on the
supply of ecosystem services
2.3.3.1. Ecosystem services. We identified ecosystem services
potentially affected by nitrogen deposition based on the generic
lists of ecosystem services published by the MA (2003) and TEEB
(2010), as well as a specific study on ecosystem services gener-
ated in Telemark (Schro €ter et al., 2014). We assume that the services
most likely to be affected by nitrogen deposition are carbon
sequestration, timber supply and biodiversity conservation. We
assume that effects on other ecosystem services identified in
Schro € ter et al. (2014) are likely to be small. For instance, snowslide
prevention will not be significantly affected by eutrophication
given that this service depends upon vegetation cover but not on
productivity or biodiversity. Tourism in Telemark is related, in
particular, to enjoying the landscape and nature of Telemark as well
as to hunting and sporting (e.g. skiing) opportunities. At current
levels of eutrophication there are not likely to be major effects on
the ecosystem components that create opportunities for recreation
such as tree cover or the presence of species that are hunted (in
particular deer and moose). We also assumed there to be no sig-
nificant effect of nitrogen deposition on the lakes of Telemark and
the services they supply (as explained above). We note that nitro-
gen deposition may potentially have a beneficial impact on farming
as nitrogen is a plant nutrient and this may reduce the need for
inorganic fertiliser application by the farmer. However atmospheric
nitrogen deposition rates are minor compared to fertiliser rate
generally applied by farmers and we do not include this in the
scope of our research (but we come back to it in the Discussion
section of our paper).
We link NEP (as assessed in step 2) to carbon sequestration and
timber production. We also analyse impacts of nitrogen on biodi-
versity in particular species diversity. For timber production, we
use annual biomass growth in stem biomass, in logged forests of
pine (Pinus sylvestris) and spruce (Picea abies) trees which account
for over 97% of timber production in Telemark (Statistics Norway,
2015). Based on De Vries et al. (1994), we derive the following
relation for impact of N deposition on stem biomass and NEP, see
Table 2 below. This is in line with Fleischer et al. (2015) who states
that the increase in wood biomass as a response to N deposition can
€ nen
be estimated at a dC/dN ratio of 19.6. It is also in line with Hyvo
et al. (2008), who shows a mean dC/dN of 25 kg N/kg C for trees and
11 kgC/kg N for soils based on 15 long-term experiments in pine
and spruce stands in Sweden and Finland.
For carbon sequestration, we only consider uptake of carbon by
forest ecosystems (which cover around 57% of the county), since
their carbon sequestration capacity is substantially larger
compared to the grasslands, croplands and glaciers covering the
rest of the county. We consider both above ground and below
ground carbon sequestration by linking carbon sequestration to the
Net Biome Production (De Vries et al., 2014) which is basically NEP
minus the loss of carbon due to fire and timber harvest. Forest fires
are very rare in Telemark (Groven and Niklasson, 2005) and carbon
losses due to forest fires are neglected in this case study. Timber
harvesting is deducted from the carbon sequestration rate, based
on the assumption that the harvested timber does not comprise a
long-term carbon storage pool. Hence, we disregard that some of
the carbon in harvested timber may be stored in wood products for
some time before returning to the atmosphere (e.g. Masera et al.,
2003).
For effects on biodiversity conservation we compare nitrogen
deposition with the critical loads per ecosystem type per EMEP grid
cell for the 20 main terrestrial ecosystem types occurring in Tele-
mark. Note that we consider biodiversity conservation to be an
ecosystem service in line with e.g. Hein et al. (2006a,b) and UN et al.
(2014) in view of the interest of society to preserve biodiversity-
rich and natural areas as a service in itself, i.e. irrespective of the
other ecosystem services such areas generate. For instance, many
societies are willing to pay to protect nature (endemic species,
natural parks) without specific consideration of the benefits this
brings in terms of (other) ecosystem services. In case critical loads
are exceeded significant impacts on biodiversity can be expected,
for example due to changes in vegetation composition (Posch et al.,
2011). However, there may even be impacts on biodiversity if
critical loads are not exceeded, in particular if loads are close to
Table 2
C:N response (kg C/kgN) for different ecosystem compartments.
Boreal needle forest Boreal broadleaved forest
Stem biomass 21.3 14.4
Coarse roots 5.0 3.1
Soil (organic þ mineral) 13.5 13.7
NEP 39.8 31.2
Source: derived from De Vries (2014)
 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663
critical loads, for instance because some plant types are more
sensitive to eutrophication than others (e.g. Grimm et al., 2008). We
therefore analyse to what extent effects on sensitive ecosystem
types occurring in Telemark have been observed at nitrogen
loading below the critical load, based on a literature review.
2.3.3.2. Adaptation. Next, it needs to be examined how ecosystem
management can be adapted in view of changes in ecosystems and
the service they supply. For instance, a logging company may adjust
harvest rates in case of significant reduction in NEP affecting the
mean annual increment in timber volume. In addition, people may
adapt through interventions in the ecosystem that counter the ef-
fects of pollution. For instance, in the Netherlands it is common
practice to regularly remove the topsoil of heathlands to reduce the
effects of eutrophication on biodiversity (Bakker and Berendse,
1999).
The societal costs of pollution are determined jointly by the
effectiveness of adaptation, the costs of adaptation and the residual
effects of pollution. In our case study, reduced atmospheric nitro-
gen deposition will lead to slower growth of trees, to which forest
managers may adapt. We assume that effects on carbon seques-
tration and on (terrestrial) biodiversity conservation do not lead to
changes in management of the forest. However, timber harvesting
regimes may be adjusted in response to changes in forest growth.
Foresters have the choice of harvesting the same amount of timber
in a shorter rotation period, or using the same rotation period and
harvesting more timber. In our scenario analysis, we assume that
the latter effect will occur, i.e. that foresters use the same rotation
period, since the rotation period not only depends upon tree
growth but also for example on the amount of labour available.
However, the economic effects of changes in nitrogen deposition on
future forestry operations also depend upon potential changes in
the sector itself. For instance, Johansen et al. (2017) identify several
options to enhance productivity and cost-effectiveness in Norwe-
gian forestry operations. Compared to our baseline scenario, for-
esters in Telemark may make better use of existing or new
technologies, and/or adjust forestry operations to deal with slower
forest growth in the future. In both cases the economic costs related
to impacts of lower nitrogen deposition on timber harvesting are
reduced. To further examine this we carry out a sensitivity analysis.
Based on Johansen et al. (2017), in our sensitivity we assume that,
for a given amount of harvested timber, a 20% decrease in costs for
the land owners is possible compared to a baseline scenario of
unchanged management.
2.3.4. Step 4. analyse economic impacts
Changes in ecosystem services supply lead to changes in wel-
fare, and it is this change e rather than the total value e that is
relevant in cost-benefit analyses (e.g. Bateman et al., 2011). Ana-
lysing the economic effects of changes in ecosystem service supply
is often not straightforward (e.g. NRC, 2005). The complexity of
valuation increases with the scale of the analysis. Changes in the
supply of ecosystem services in a large area may lead to price ef-
fects, in particular for provisioning services. In such cases, supply
and demand curves for the good in question have to be considered.
Regulating services are generally more complex to analyse because
there is a lack of market information from which supply and de-
mand curves can be derived. For cultural services, economic effects
on tourism-related ecosystem services are often relatively easy to
assess given that many aspects of tourism are recorded in market
statistics (e.g. overnight stays in hotels, restaurant visits), whereas
the travel cost method provides an entry point for analysing the
consumer surplus related to tourism and recreation. Other cultural
services, e.g. the appreciation of the nature and tranquillity of the
Telemark landscape, are harder to value.
655
In our case study, we analyse the changes in economic surplus
generated by changes in the supply of timber and in carbon
sequestration as a consequence of changes in nitrogen deposition.
For timber, we use detailed statistics on the Norwegian timber
market (with much of the data disaggregated by county) provided
by Statistics Norway, in order to estimate the value of production
and the likelihood of price changes as a consequence of changes in
N deposition rates over time.
For carbon, we use an avoided damage cost approach. We as-
sume the marginal value of one unit of carbon sequestered is
constant because carbon sequestration in Telemark will not affect
the (global) marginal value of one unit of carbon sequestered. The
use of the avoided damage cost approach is in line with other work
in this field and reflects the difficulties of analysing and discounting
the future welfare effects of current changes in carbon emissions
(e.g. Stern, 2008). We do not monetise effects on biodiversity given
the methodological uncertainties surrounding such an analysis (e.g.
Cummings and Harrison, 1995; Spash, 2008). All values and prices
are expressed in V2010, with an exchange rate of 8 NOK for the euro.
3. Results
3.1. Nitrogen deposition in Telemark under different scenarios
Table 3 presents the nitrogen deposition rates, including oxi-
dised and reduced nitrogen in Telemark in the baseline and four
future scenarios. The values are averaged across the 29 EMEP grid
cells (shown in Annex 1) and across ecosystem types. The table
shows that average nitrogen deposition rates are forecasted to
decline, from the present 3.82 kgN/ha/year to 2.77 kgN/ha/year in
the OPT2030 scenario.
3.2. Analysing the effects on the ecosystems
For forest ecosystems, the relation between N deposition and
NEP is well understood (De Vries et al., 2014). However, for other
ecosystems there are fewer studies linking N deposition and C
uptake, and uncertainty is much higher. For UK heathlands, for
example, Evans et al. (2006) and De Vries et al. (2009) report that
above ground biomass responses were in the range of 5e15 kgC/kg
N, but only 7 measurements were available, for three sites. In the
same sites, below ground biomass uptake ranged from
12 to
34 kg C/kgN (with an average of 22 kg C/kg N). However, extrap-
olation of so few data from UK to Telemark Norway cannot be
justified. Also for natural grasslands there are very few studies
linking N deposition and C uptake, including in Nordic countries.
We therefore follow a different approach to provide some first level
understanding of the productivity response of non-forest ecosys-
tems in Telemark to N deposition. An elaborate, in-depth study by
De Wit et al. (2015) has analysed carbon fluxes in Norwegian
terrestrial (and aquatic) ecosystems. This study found that carbon
uptake in Norwegian mountain vegetation biomass is 5 ± 3 kg C/ha/
year. We assume that 5 kg C/ha/year represents the carbon flux in
Telemark for the ecosystem types ‘Alpine heath communities’,
‘Alpine meadow communities’ and ‘Snow bed vegetation’. This
carbon flux represents the upper boundary of the response of the
vegetation to N deposition. This response would be around 5 kg C/
ha/year if the complete carbon flux could be attributed to the effect
of N deposition. This, of course, is not the case, hence the 5 kg C/ha/
year presents the upper boundary. The actual response of the alpine
and snow bed vegetation to N deposition is somewhere between
0 and 5 kg C/ha/year.
The carbon balance in peatlands is strongly dependent upon the
drainage status: drained peatland will emit carbon, undrained
peatlands may sequester carbon. De Wit et al. (2015) estimate that
656 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663

Table 3
Average annual total nitrogen depositions per scenario in kg N per hectare.

Average Nitrogen deposition (over Telemark) 2010 CLE_ 2020 Goth 2020 CLE 2030 OPT 2030

Total nitrogen 3.82 3.22 3.33 2.98 2.77

Wet deposition NOx 1.71 1.26 1.33 1.04 1.02
Dry deposition NOx 0.53 0.41 0.45 0.35 0.35
Wet deposition NH3 1.19 1.17 1.20 1.19 1.03
Dry deposition NH3 0.36 0.37 0.36 0.40 0.38

around 15% of peatlands are drained in Norway south of Finnmark and furthermore consider that 34% of the additional increment in
(subarctic Norway). The net effect of nitrogen deposition is difficult stem biomass due to nitrogen deposition is logged (based on Sta-
to estimate. On the one hand increased nitrogen levels may in- tistics Norway data showing that on average 34% of the mean
crease plant growth, but on the other hand it may enhance the annual increment in the forests of Telemark is logged, on an annual
activity of heterotrophic bacteria increasing carbon emissions from basis). The results are shown in Table 5, for the 2010 baseline and
peat oxidation. Peatland including fens and bogs cover 3% of Tele- each of the four future scenarios. Note that we do not consider the
mark, of which around 60% in the area on or around the Hardan- effects of temperature changes in the future scenarios (but we
garvida plateau where N deposition rates are relatively low (EMEP come back to this in the Discussion section). Fig. 2 presents the
grid cells 1 to 13). Given the uncertainties involved we do not spatial pattern of timber logging of stem biomass dependent upon
further consider the effects of N deposition on NEP (and carbon nitrogen deposition.
sequestration) on peatlands.
Table 4 shows that nitrogen deposition N affects, in particular,
the NEP of forest ecosystems (with associated impacts on carbon 3.3.2. Carbon sequestration
sequestration and timber production). This is consistent with, for The impacts on carbon sequestered in the ecosystem is deter-
example, De Vries et al. (2014) who states that forest ecosystems mined by the effect of nitrogen deposition on stem, branch and
are the main carbon sink in Europe. Note that effects on biodiversity coarse roots biomass, i.e. the long-term organic carbon pools, as
are more subtle and not only depend upon ecosystem productivity; well as on soil carbon. Accumulation of soil carbon occurs due to
some species may be particularly vulnerable and shifts in species increased litter rates (De Vries et al., 2014). The sum of tree and soil
composition may occur even at low N deposition rates (Bobbink C sequestration equals the NEP. However, the storage in the
and Hettelingh, 2011). ecosystem, the net biome production, is also influenced by carbon
losses due to fire and timber harvest. As explained above forest fires
3.3. Analysing effects on ecosystem services are relatively rare in Telemark (Groven and Niklasson, 2005) and
are disregarded. As explained above, timber harvesting is deducted
3.3.1. Timber production from the carbon sequestration rate. This is a simplification, since
Timber production is influenced by the acidifying and eutro- some of the timber may be used for house or furniture construction
phying effects of nitrogen deposition. We assume that all wood and thereby be removed from the atmosphere for an extended
harvested in Telemark is logged in coniferous forests (pine and period of time. Fig. 3 presents carbon sequestration in Telemark
spruce forests accounted for over 97% of all wood harvest in Tele- under different scenarios.
mark by volume, Statistics Norway, 2015), that no logging takes
place in the (sparse) forests in EMEP grid cells 1 to 3 (middle of the
Hardanger plateau) and that in the remaining parts of Telemark Table 5
96% of all forests are logged (Statistics Norway, 2015). Converting Effects of N deposition on ecosystem services (physical units).
NEP to timber production requires understanding the allocation of Scenario Timber supply Carbon sequestration
plant biomass to harvestable wood. We assume that all stem
m3 timber/year ton CO2/year
biomass can be harvested (smaller parts of the stem are suitable for
pulping), and that the C:N response of the stem biomass is Baseline 2010 121961 431903
TSAP CLE 2020 103767 367307
21.3 kg C/kg N (De Vries et al., 2014). Furthermore, it is assumed
Gothenburg Prot. 2020 107352 380203
that dry weight has a carbon content of 50% and that the basic WPE TSAP CLE 2030 96224 340206
wood density is 471 kg m
3 (Masera et al., 2003). We calculate the TSAP OPT 2030 89054 314873
effect of nitrogen deposition per EMEP grid cell on timber forest,

Table 4
Impacts of N deposition on NEP (in kg C/ha/yr), different scenarios.

Baseline CLE 2020 Got2020 CLE2030 OPT2030

Pine þ Spruce forest 156 133 137 0 114

Deciduous birch forest 134 114 118 0 98
Thermophilic deciduous forest 139 118 122 0 101
Alpine heath communities <5 <5 <5 <5 <5
Alpine meadow communities <5 <5 <5 <5 <5
Peatlands (fens and bogs) pma pm pm pm pm
Snow bed vegetation <5 <5 <5 <5 <5
a
pm ¼ pro memorie: effects of N deposition on NEP in this ecosystem has not been considered, as motivated above.
 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663 657

Fig. 2. Map of timber logging dependency on N deposition in Telemark under different scenarios.

Fig. 3. Map of carbon sequestration in Telemark under different scenarios. The left hand figure presents carbon sequestration in the 2010 baseline.
658 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663
3.3.3. Biodiversity conservation
To assess impacts of eutrophication on biodiversity conservation
we first compare actual deposition rates in the scenarios with the
critical loads, by ecosystem. Annex 2 presents the critical loads for
the ecosystem types found in Telemark, as derived from Bobbink
and Hettelingh (2011). The Annex shows that the critical load for
nitrogen loading is not surpassed in any of the terrestrial ecosys-
tems. Nitrogen deposition in the coastal areas of Telemark is
highest, i.e. between 5 and 6 kg N/ha/year in the 2010 baseline,
which is below the critical loads of the natural ecosystems found in
this part of the county (coniferous, mixed and broadleaved forests),
i.e. around 10e15 kg N/ha/year. The critical loads for the most
sensitive ecosystems, (‘Grass and dwarf willow snow patch’ and
‘Poor bryophyte snow patch’) are much lower than for forest eco-
systems, i.e. around 3e5 kg N/ha/year. The most sensitive alpine
ecosystems are found in the uplands of Telemark where nitrogen
deposition rates are between 1 and 2 kg N/ha/year in 2010. Natural
nitrogen deposition loads, in comparison, are in the order of
0.5 kg N/ha/year (Stevens et al., 2010).
In general, even though critical loads are not surpassed and
there are no major chemical transformations in the ecosystem,
there may still be effects on plant communities (Grimm et al.,
2008). Responses to nitrogen loading may include shifts in domi-
nant plant species including increased invasiveness or dominance
of N-demanding species. Given that arctic grasslands are particu-
larly sensitive to nitrogen loading (Shaver and Chapin, 1995), and
that critical loads are approached in these ecosystems in Telemark,
we focus our analysis on this ecosystem type, recognising that
vegetation responses to nitrogen deposition are highly complex
and cannot be treated in detail in this paper (but see Bobbink et al.,
1998 for an example of a more elaborate analysis). For instance, the
effects of nitrogen deposition and acidification in grasslands are
interlinked: acidification may prevent plants using the additional
nutrients made available through nitrogen deposition (e.g. Stevens
et al., 2010). There is a scarcity of data on vegetation changes in
Telemark, but Klanderud and Birks (2003) examine a long-term
time series in the Jotunheimen mountains, just north of Tele-
mark. They find that increased deposition of nitrogen and changes
in grazing and tourism might have influenced some of the species
turnovers, but recent climatic changes are the most likely major
driving factor for the changes observed (Klanderud and Birks,
2003). In this paper we are concerned with the effects of the four
policy scenarios on biodiversity, with nitrogen deposition on alpine
heath and grasslands in Telemark being reduced to 0.8 to 1.5 in the
2030 OPT scenario that involves the highest emission reduction.
Based on Klanderud and Birks (2003), Bobbink and Hettelingh
(2011) and Stevens et al. (2010) we assume that changes in nitro-
gen deposition (a reduction of around 0.5 kg N/ha/year in the 2030
OPT scenario) as a function of the four policy scenarios are unlikely
to have a major effect on biodiversity in Telemark. Our assessment
also seems justified when compared with work in the US. For
instance, Bowman et al. (2006) find that, in alpine ecosystems in
the US Rocky Mountains, the critical load for changes in individual
species is around 4 kg N ha
1 yr
1 and for overall community
change this is around 10 kg N ha
1 yr
1.
Table 5 summarises effects of nitrogen deposition on all three
examined ecosystem services. As for the service ‘Biodiversity con-
servation’, there are no areas where the critical load is exceeded in
Telemark's ecosystems in the baseline as well as in the four
examined scenarios.
3.3.4. Step 4. analyse changes in the value generated by ecosystems
In step 4, potential shifts in demand and supply curves for
timber and, consequently, price effects need to be considered.
Telemark provides around 6% of the Norwegian timber (by volume,
in 2015, Statistics Norway, 2015). This means that for this service
price effects of change in supply cannot be a priori disregarded.
Around half of Norwegian roundwood harvest is sold for pulpwood
to the paper industry, and the other half is used for other purposes
in particular construction and firewood (Bolkesjo €, 2005). Pulping
depends mainly upon spruce wood although pine wood is also
used. Pulping is the most important market for timber produced in
Telemark. Prices for pulpwood have decreased in the period
2010e2015 by around 10% depending upon species and use of the
wood, with pulpwood prices facing the largest declines (Statistics
Norway, 2015), at a roughly constant volume of timber. Two main
factors driving timber price changes in Norway including Telemark
are the availability of wood (which has been increasing over time
due to low harvest/regrowth rates (see the next section) and in-
vestments in the pulp sector. Bolkesjo € (2005) presents the results of
a Computed General Equilibrium (CGE) model for pulpwood in
Norway and finds that major increases in demand by increasing the
capacity of paper mills (or major reductions in supply e.g. due to
decreased nitrogen loading) will only have a small effect on wood
prices, partly because of the possibility to substitute Norwegian
timber with timber from other countries. Therefore, we assume no
effect of reduced nitrogen deposition rates and reduced timber
growth on timber prices. We also assume that, in line with current
practices, reductions in timber harvest will be compensated by
increases in timber imports. Consequently, there will be no change
in the associated consumer surplus. We therefore express the
economic value of timber production in terms of changes in pro-
ducer surplus. We approximate the producer surplus that can be
attributed to the change in timber production by the changes in
profits that forest owners make when they sell the wood (se e.g.
Varian, 1992 for details). We assume that the changes in timber
productivity do not change the fixed costs of harvesting, i.e. the
owner still needs equipment, own the land, etc. However the var-
iable costs will change as a function of changes in harvest volume.
In 2010, wood harvest in Telemark consist of 62% spruce and 38%
pine, with road-side prices of respectively 42 and 45 euro/m3
(Statistics Norway, 2015). Average road side prices for coniferous
woods are therefore 43 euro/m3, in 2010. We assume silviculture
based on natural regeneration and low land preparation costs
excluding subsidies of 125 euro/ha (Tromborg and Sjolie, 2011)
with a rotation period of 75 years on average for spruce and pine
forests (Follo, 2011). This translates into a planting costs of 1.4 euro/
m3. Harvest costs in Telemark are 18 euro/m3 (Tromborg and Sjolie,
2011). Hence, the total variable costs are 19 euro/m3, and the dif-
ference between the roadside price and the variable costs amounts
to 43-19 ¼ 24 euro/m3. Based on the assumptions motivated above,
this value can be multiplied with the annual timber harvest under
different scenarios as specified in Table 5. Table 6 specifies that
compared to 2010 the economic value generated by timber pro-
duction is reduced in each of the scenarios, from 2.9 million euro
per year in 2010 to 2.1 million euro per year in the OPT 2030
scenario.
For carbon, we use an avoided damage cost approach. We use a
carbon price of 30 euro per ton CO2 as used as baseline in EU level
policy analyses (EC, 2009). This is comparable with the social costs
of carbon proposed by the US EPA (at a 3% social discount rate). The
damage cost estimate is also in line with some of the modelling
studies on the costs of climate change (e.g. Nordhaus, 2013) even
though the uncertainty is high and other, higher estimates are also
available in the literature (e.g. Van den Bergh and Botzen, 2014). We
come back to this in the discussion section. As mentioned above we
 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663 659

Table 6
Effects of N deposition on ecosystem services (monetary units).

Scenario Timber supply Carbon sequestration

Gross revenue (million euro/year) Value (producer surplus) Avoided damage costs (million euro/year)
(million euro/year)

Baseline 2010 5.3 2.9 13.0

TSAP CLE 2020 4.5 2.5 11.0
Gothenburg Prot. 2020 4.6 2.6 11.4
WPE TSAP CLE 2030 4.1 2.3 10.2
TSAP OPT 2030 3.8 2.1 9.4

do not value changes in biodiversity in monetary terms. Table 6 function of uncertainties in the preceding steps, as well as in un-
presents results for timber supply and carbon sequestration. certainty related to how people respond to changes in ecosystem
services supply. For example, in this study it was assumed that
forest management practices do not change as a consequence of
4. Discussion
reductions in nitrogen deposition. Such assumptions should be
made explicit and motivated in studies analysing air pollution ef-
4.1. Key sources of uncertainty and sensitivity analysis
fects on ecosystem services. In addition, temperature is a main
factor influencing forests in Telemark. In scenario analysis, the
In the first step of the analysis there are uncertainties related to
combined effects of eutrophication and climate change would
the deposition of pollutants across the landscape and over time. In
ideally be assessed in an integrated manner. Climate change may
the EMEP model used to estimate annual deposition rates, temporal
affect the three examined services carbon sequestration, timber
variations are considered by running the model for a five years
production, biodiversity conservation but potentially also bring
period and taking running averages. However, there remain un-
other effects such as changes in snowslide risks or erosion due to
certainties in deposition rates as a function of uncertainties in
changes in rainfall patterns.
meteorological conditions, in the actual emissions of pollutants and
Finally, there are a range of factors of uncertainty related to
in chemical reactions in the atmosphere, among others (see
valuing changes in ecosystem services supply. First, it is hard to
Simpson et al., 2012 for details). Over a time frame of decades,
predict future prices of ecosystem services, even when they are
climate change may affect prevailing temperatures or wind di-
traded on a market (e.g. Piot-Lepetit, and M'Barek, 2011). This is
rections which further contributes to uncertainty in deposition
related to the difficulty of forecasting both future demand as a
rates.
function of changes in preferences for the service involved and the
In the second step, analysing ecosystem responses, un-
entry of potential substitutes on the market, and to forecasting
certainties in dose-response relationships (e.g. linking nitrogen
future supply as a function of among others land use and climate
deposition to changes in NEP) need to be considered. There are
change. To analyse effects of air pollution, potential changes in the
several sources of uncertainty in these relationships. First, air pol-
supply and demand curve for the service involved need to be
lutants can have important indirect effects. Indirect effects include
considered, as illustrated in this paper for timber supply in Tele-
those in which the pollutant alters the physical or chemical envi-
mark. If these changes are significant, a Computed General Equi-
ronment (e.g., soil properties), the plant's ability to compete for
librium (CGE) modelling approach is needed to investigate how
limited resources (e.g., water, light), or the plant's ability to with-
markets might change as a function of air pollution policies. CGE
stand pests or pathogens (Bowman et al., 2006). Indirect effects are
modelling requires estimating a range of parameters including in-
often difficult to observe yet may have substantial impacts on dose-
puts and outputs of economic sectors, substitution elasticities be-
response relations (Smith, 1990). Second, ecosystem responses
tween inputs in these sectors and linkages between environment
often depend upon a combination of pollutants and other factors
and economic production, and dealing with associated un-
(Brunner and Clark, 1997). For instance, N deposition may have a
certainties (e.g. Abler et al., 1999). This is particularly relevant when
favourable effect on ecosystem productivity, but other types of air
analysis of air pollution effects on ecosystem services are scaled up
pollution including O3 and SO2 will generally affect productivity
to larger areas (e.g. Europe), when significant price effects of
negatively (EPA, 1999), see also IIASA (2016). There may also be a
changes in pollution on ecosystem services are much more likely.
strong interaction between acidification and eutrophication as
We conducted a sensitivity analysis for one of our key as-
drivers for ecosystem change, with soil acidity having a major
sumptions, i.e. that forest management doesn't change when
impact on soil nutrient availability (e.g. Stevens et al., 2010). Over
reduced eutrophication leads to slower growth of trees. In our
time, climate change may influence local temperatures and water
sensitivity analysis we assume that costs can be decreased by 20%
availability which may change the resilience of the ecosystem to
compared to the baseline through adjustment of forestry opera-
cope with the deposition of air pollutants (Reich et al., 2006).
tions including adaptation to lower tree growth (based on Johansen
Finally, lag effects may occur in the response of ecosystems to in-
et al., 2017). This reduction in costs has significant ramifications for
creases or reductions in pollution loading. This happens for
the producer surplus generated by the timber supply service in the
example when pollutants or substances in the soil formed due to
four scenarios, which increases to 2.9, 3.0, 2.7, and 2.5 million euro
pollution are buffered, as in the case of the various hydrogen (Hþ)
per year respectively for the TSAP CLE 2020, Gothenburg Prot. 2020,
buffers in the soil. Effects of pollutants can therefore continue after
WPE TSAP CLE 2030 and TSAP OPT 2030 scenarios. In other words, a
the stressors themselves have been reduced. This effect has not
substantial part of the reduction in producer surplus due to changes
been considered in this study, but lag effects may delay the effects
in air pollution can be mitigated by enhancing the efficiency of
of air pollution policies on ecosystem services.
forestry operations. This result reflects that the uncertainties in this
In the third step, uncertainties in changes in ecosystem services
kind of scenario analyses are high, caused by the sensitivity of the
supply need to be considered. These uncertainties occur as a
660 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663
analysis to key assumptions (cf. e.g. Moiseyev et al., 2011;
Papadimitriou, 2012). It also confirms the need for sensitivity an-
alyses when CBA is used in support of policy making on air pollu-
tion reductions.
4.2. Implications for supporting air pollution policies
The case study for Telemark shows the potential as well as the
challenges of linking ecosystem services to air pollution in support
of analysing costs and benefits of different policy scenarios. First
the paper shows that changes in air pollution may entail costs as
well as benefits depending upon the pollutant and service involved.
However, the effects of different air pollutants on ecosystems are
entirely dependent upon the type of pollutant. For instance, the
effects of PM2.5 on ecosystems are probably limited (the main ef-
fects of PM2.5 are being related to increasing health risks related to
cardiovascular and respiratory diseases). SOx deposition continues
to affect Nordic ecosystems through acidification of soils, lakes and
streams, even though this deposition is much lower when
compared to the 1980s. This entails costs for society, related to
liming of lakes and streams as well as to residual impacts on eco-
systems. Clearly, it cannot be assumed that the analysis we con-
ducted is representative, and the full mix of pollutants and their
effects needs to be considered in policy analysis.
Second, the paper shows that detailed spatial models are
required to link air pollution and ecosystem impacts. In terms of
deposition, the resolution of the EMEP grid is consistent with the
needs of such analyses. The 28  28 km size of the cells means that
deposition in different ecosystem types can be modelled e in
Telemark this is highly relevant because the sensitive meadows and
heathlands located in the upland parts of the county received a
lower N deposition. From other studies (e.g. Schro € ter et al., 2014;
Remme et al., 2015; Zank et al., 2016) it is known that to account
for heterogeneity of the landscape a spatial resolution as small as
100  100 m may be needed to model ecosystem state and
ecosystem service flows e depending upon the heterogeneity of
the landscape. Hence, in scaling up, major spatial datasets and
substantial computing power are needed to link depositions and
ecosystem service impacts at national or continental scales.
Importantly, third, it needs to be considered in scaling up how
additional ecosystem services and their values can be included in
the scenario analyses (Smart et al., 2011). It is noteworthy that the
increasing application of ecosystem accounting is likely to pro-
gressively provide baseline data on ecosystem services for scenario
analyses. Ecosystem accounting involves recording ecosystem
condition, services and values generated by these services in an
accounting framework, based on detailed maps of ecosystems and
the services they supply (Hein et al., 2015). It is now being tested at
(sub-)national scales (e.g. De Jong et al., 2015; Statistics Norway,
2015) and at the European scale (e.g. Maes et al., 2016; Ledoux
and Wejchert, 2017).
CBAs of air pollution policies may be extended with an analysis
of stakeholders losing and gaining in different policy scenarios, in
order to inform policy makers. Basically, following Fig. 1, this in-
volves connecting changes in ecosystem services supply to specific
stakeholders (see e.g. Suwarno et al., 2016). In the case of Telemark,
for example, forest owners involved in timber harvesting benefit
from eutrophication, and society at large benefits from enhanced
carbon sequestration due to eutrophication.
An important question is to what degree an analysis of
ecosystem services will change the outcomes of CBAs of air pollu-
tion control policies, compared to current methodologies.
Currently, ecosystem services impacts are not or only to a very
limited degree (e.g. only crop damages) considered in these CBAs
(e.g. IIASA, 2014), and the majority of the benefits of air pollution
control are derived from statistical reductions in premature mor-
tality valued based on the Value of a Life Year (VOLY) (e.g. Hein
et al., 2016a,b). Further testing is required, including a larger
array of ecosystem services and pollutants, in order to pinpoint the
relative importance of health versus ecosystem service-related
benefits of cleaner air (e.g. by analysing both health and
ecosystem services related benefits for specific areas). As analysed
in this paper, both positive and negative effects of policies on
ecosystem services need to be considered.
5. Conclusions
This study presents a general approach for connecting air
pollution to ecosystem services supply, and a case study for Tele-
mark county, Norway. We show that, in this county, timber pro-
duction and carbon sequestration benefit from nitrogen deposition.
The avoided damage costs due to carbon sequestration as a function
of nitrogen deposition in Telemark in 2010 are estimated at 13
million euro per year, and the value of the additional timber har-
vesting due to nitrogen deposition in the same year is estimated at
2.9 million euro. With the nitrogen deposition reduction assumed
in four policy scenarios, these benefits would be reduced, down to
respectively 9.4 and 2.1 million euro in 2030 in the Optimal sce-
nario. We did not find a negative effect of nitrogen deposition on
biodiversity conservation at current deposition rates in Telemark
but cannot exclude such an effect with certainty given the limita-
tions of our approach (focussing only on the main ecosystem types
in Telemark).
We only analysed the impacts of dry and wet nitrogen deposi-
tion and want to reiterate that other pollutants may have different
effects on ecosystem state and on ecosystem services supply (for
instance the deposition of SOx is still leading to acidification of
Norwegian lakes and streams including in Telemark). A costs-
benefit analysis of the effects of air pollution scenarios on
ecosystem services would need to consider all pollutants, and their
impacts on a broad set of ecosystem services. Climate change and
air pollutant exposure may mutually reinforce or reduce specific
aspects of ecosystem change.
Our study shows that there are significant challenges in scaling
up, which would be required if this kind of analyses is used to
support air quality policy making at the European scale. These
challenges include dealing with spatial and temporal variability in
ecosystem state and services supply, uncertainties in dose-
response relationships and value estimates, specifying adaptation
options and their impacts, and dealing with complex ecosystem
dynamics (thresholds, lag effects) and combined effects of multiple
stressors. These key sources of uncertainty need to be better un-
derstood and more case studies are needed to broaden insights in
the effects of air pollution on ecosystem services. With these con-
siderations an ecosystem services approach could be applied to
give quantitative support to air pollution policy making at national
or continental scale.
Acknowledgements
Confidence Duku and Bart Steen are thanked for their assistance
€ ter is thanked for making data on
with GIS analyses. Matthias Schro
Telemark available while working on his PhD studies at Wagenin-
gen University. Robin Nelson and Vikram Paul are thanked for their
comments. We also thank two anonymous reviewers for their very
helpful comments.
 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663 661

Annex 1. EMEP grids superimposed on Telemark.

Annex 2. Critical loads for the different ecosystem types

found in Telemark (terrestrial ecosystems only).

Ecosystem type Critical load (range, average) in kg N/ha/yr N loading per ecosystem type per scenario
(average across EMEP grid cells in Telemark)

2010 baseline CLE 2020 Gothenburg 2020 CLE 2030 OPT 2030

Coniferous forest (dense) 10-15, 12.5 3.9 3.3 3.4 3.1 2.8
Coniferous and mixed forest (open) 10-15, 12.5 3.9 3.3 3.4 3.1 2.9
Lichen rich pine forest 5-10, 7.5 4.2 3.6 3.7 3.3 3.1
Low herb broadleaved forest 10-15, 12.5 3.4 2.9 3.0 2.7 2.5
Tall-fern and tall-herb broadleaved forest 10-15, 12.5 3.7 3.1 3.3 2.9 2.7
Bilberry birch forest 5-8, 6.5 3.3 2.8 2.9 2.6 2.4
Cowberry birch forest 5-8, 6.5 3.4 2.9 3.0 2.7 2.5
Lichen rich birch forest 5-8, 6.5 3.8 3.2 3.3 3.0 2.7
Ombrotrophic hummock and lawn bog 5-10, 7.5 2.4 2.0 2.1 1.8 1.7
Rich lawn fen 10-15, 12.5 2.5 2.1 2.2 1.9 1.8
Rich mud-bottom fen 10-15, 12.5 3.2 2.7 2.8 2.5 2.3
Alpine ridge vegetation and barren land 5-15, 10 2.0 1.7 1.7 1.5 1.4
Graminoid and wood-rush ridge 10-20, 15 1.7 1.4 1.5 1.3 1.2
Heather rich alpine ridge vegetation 10-20, 15 1.8 1.5 1.6 1.4 1.3
Lichen rich alpine ridge vegetation 5-15, 10 1.7 1.4 1.5 1.3 1.2
Early snow patch vegetation 5-15, 10 2.0 1.7 1.8 1.6 1.5
Alpine heather and dwarf birch heath 5-15, 10 2.1 1.8 1.8 1.6 1.5
Alpine fern meadow 5-15, 10 2.1 1.8 1.8 1.6 1.5
Grass and dwarf willow snow patch 3-5, 4 1.7 1.4 1.5 1.3 1.2
Poor bryophyte snow patch 3-5, 4 1.9 1.7 1.7 1.5 1.4
662 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663

References Follo, G., 2011. Factors influencing Norwegian small-scale private forest owners'
ability to meet the political goals. Scand. J. For. Res. 26 (4), 385e393.
Grimm, N.B., Foster, D., Groffman, P., Grove, J.M., Hopkinson, C.S., et al., 2008. The
Aber, J., McDowell, William, Nadelhoffer, Knute, Magill, Alison, Berntson, Glenn,
changing landscape: ecosystem responses to urbanization and pollution across
Kamakea, Mark, McNulty, Steven, Currie, William, Rustad, Lindsey,
climatic and societal gradients. Front. Ecol. Environ. 2008 (6), 264e272.
Fernandez, Ivan, 1998. Nitrogen saturation in temperate forest ecosystems.
Groven, R., Niklasson, M., 2005. Anthropogenic impact on past and present fire
BioScience 48 (11), 921e934. Nov., 1998.
regimes in a boreal forest landscape of southeastern Norway. Can. J. For. Res. 35,
Abler, D.G., Rodríguez, A.G., Shortle, J.S., 1999. Parameter uncertainty in CGE
2719e2726, 2005.
modeling of the environmental impacts of economic policies. Environ. Resour.
Haines-Young, R., Potschin, M., 2012. Common International Classification of
Econ. 14 (Issue 1), 75e94.
€glund-Isaksson, L., Ecosystem Services (CICES, Version 4.1), vol. 33. European Environment Agency.
Amann, M., Bertok, I., Borken-Kleefeld, J., Cofala, J., Heyes, C., Ho
Hein, L.C., Obst, B., Edens, Remme, R.P., 2015. Progress and challenges in the
Klimont, Z., Nguyen, B., Posch, M., Rafaj, P., Sandler, R., 2011. Cost-effective
development of ecosystem accounting as a tool to analyse ecosystem capital.
control of air quality and greenhouse gases in Europe: modeling and policy
Curr. Opin. Environ. Sustain. 14, 86e92.
applications. Environ. Model. Software 26 (12), 1489e1501.
Hein, L., Bagstad, K., Edens, B., Obst, C., de Jong, R., Lesschen, J.P., 2016a. Defining
Austnes, K., 2013. Vurdering Av Fortsatt Kalkingsbehov I Kalkede Innsjøer I Tele-
ecosystem assets for natural capital accounting. PLos One 11 (11), e0164460.
mark. Research report. Series/Report: NIVA-rapport; 6507.
https://doi.org/10.1371/journal.pone.0164460.
Bakker, J.P., Berendse, F., 1999. Constraints in the restoration of ecological diversity
Hein, L., Roberts, P., Gonzalez, L., 2016b. Valuing a statistical Life year in relation to
in grassland and heathland communities. Trends Ecol. Evol. 14 (2), 63e68.
clean air. J. Environ. Assess. Pol. Manag. 18, 16e25.
Bateman, I.J., Mace, G.M., Fezzi, C., Atkinson, G., Turner, K., 2011. Economic analysis
Hettelingh, J.P., Posch, M., Slootweg, J., Reinds, G.J., Spranger, T., Tarrason, L., 2007.
for ecosystem service assessments. Environ. Resour. Econ. 48 (2), 177e218.
Critical loads and dynamic modelling to assess european areas at risk of acid-
Bechmann, M.E.D., Berge, H.O., Eggestad, S.M., Vandsemb, 2005. Phosphorus
ification and eutrophication. Water Air Soil Pollut. 7, 379e384.
transfer from agricultural areas and its impact on the eutrophication of lake-
Hogberg, P., Fan, H., Quis, M., Binkley, D., Tamm, C.O., 2006. Tree growth and soil
sdtwo long-term integrated studies from Norway. J. Hydrol. 304 (1e4),
acidification in response to 30 years of experimental nitrogen loading on boreal
238e250, 10 March 2005.
€gberg, P., 1997. Does atmospheric deposition of nitrogen threaten forest. Global Change Biol. 12, 489e499, 2006.
Binkley, D., Ho
Hyvo€ nen, R., Persson, T., Andersson, S., Olsson, B., Ågren, G.I., Linder, S., 2008.
Swedish forests? Forest Ecol. Manag. 92, 119e152.
Impact of long-term nitrogen addition on carbon stocks in trees and soils in
Bobbink, R., Horning, M., Roelofs, J.G.M., 1998. The effects of air-borne nitrogen
northern Europe. Biogeochemistry 89, 121e137.
pollutants on species diversity in natural and semi-natural European vegeta-
International Institute for Applied Systems Analysis (IIASA), 2014. IIASA TSAP
tion. J. Ecol. 86 (5), 717e738.
Report 11; Cost-benefit Analysis of Final Policy Scenarios for the EU Clean Air
Bobbink, R., Hettelingh, J.P., 2011. In: Review and Revision of Empirical Critical Loads
Package. Version 2a. IIASA, Laxenburg, Austria. October 2014.
and Dose-response Relationships. Coordination Centre for Effects, National
International Institute for Applied Systems Analysis (IIASA), 2016. Thematic Strat-
Institute for Public Health and the Environment (RIVM), the Netherlands.
egy on Air Pollution (TSAP) Report #16, Working Party on the Environment
Bolkesjø, T.F., 2005. Projecting pulpwood prices under different assumptions on
(WPE) for the European Council under 2014 Current Legislation (CLE) for 2030
future capacities in the pulp and paper industry. Silva Fenn 39, 103e116.
Using the PRIMES 2013 Reference Activity Projection. IIASA, Laxenburg, Austria.
Bowman, W.D., Gartner, J.R., Holland, K., Wiedermann, M., 2006. Nitrogen critical
Johansen, U., Werner, A., Nørstebø, V., 2017. Optimizing the wood value chain in
loads for alpine vegetation and terrestrial ecosystem response: are we there
northern Norway taking into account national and regional economic trade-
yet? Ecol. Appl. 16, 1183e1193.
offs. Forests 8, 172e193, 2017.
Brunner, R.D., Clark, T.W., 1997. A practice-based approach to ecosystem manage-
Jones, L., Mills, G., Milne, A., 2014. Assessment of the Impacts of Air Pollution on
ment. Conserv. Biol. 11, 48e58. https://doi.org/10.1046/j.1523-
Ecosystem Services e Gap Filling and Research Recommendations. Centre for
1739.1997.96005.x.
Ecology and Hydrology, Bangor UK. Final Report.
Capros, P., Mantzos, L., Parousos, L., Tasios, N., Klaassen, G., Van Ierland, T., 2011.
Klanderud, K., Birks, H.J.B., 2003. Recent increases in species richness and shifts in
Analysis of the EU policy package on climate change and renewables. Energy
altitudinal distributions of Norwegian mountain plants. Holocene 13 (1), 1e6,
Pol. 39 (3), 1476e1485.
2003.
Cummings, R.G., Harrison, G.W., 1995. The measurement and decomposition of
Ledoux, L., Wejchert, J., 2017. 2e17. Developing Pilot Ecosystem Accounts in the
nonuse values: a critical review. Environ. Resour. Econ. 5, 225e247.
European Union: Potential Policy Applications. Chapter 23 in World Bank.
Daily, G.C., Polasky, S., Goldstein, J., Kareiva, P.M., Mooney, H.A., Pejchar, L.,
Better Policy through Natural Capital Accounting: Stocktaking and Ways For-
Ricketts, T.H., Salzman, J., Shallenberger, R., 2009. Ecosystem services in deci-
ward. World Bank, Washington DC.
sion making: time to deliver. Front. Ecol. Environ. 7, 21e28.
MA, 2003. Millennium Ecosystem Assessment. Ecosystems and Human Well-being:
De Jong, R., Edens, B., Van Leeuwen, N., Schenau, S., Remme, R., Hein, L., 2015.
a Framework for Assessment. Report of the Conceptual Framework Working
Ecosystem Accounting in Limburg Province, the Netherlands; Part I: Physical
Group of the Millennium Ecosystem Assessment. Island Press, Washington, DC.
Supply and Condition Accounts. Statistics Netherlands (CBS) and Wageningen
MA, 2005. Millennium Ecosystem Assessment. Ecosystems and Human Well-being:
University.
Global Assessment Report, 1 to 3. Island Press, Washington, DC.
De Smet, L., Devoldere, K., Vermoote, S., 2007. Valuation of Air Pollution Ecosystem
Maes, J.B., Egoh, L., Willemen, C., Liquete, P., Vihervaara, Sch€agner, J.P., Grizzetti, B.,
Damage, Acid Rain, Ozone, Nitrogen and Biodiversity. ENV.C.5/SER/2006/0114r.
Drakou, E.G., Notte, A.L., Zulian, G., Bouraoui, F., Luisa Paracchini, M., Braat, L.,
European Commission, DG ENV. 06/11867/SV. Arcadis Ecolas.
Bidoglio, G., 2012. Mapping ecosystem services for policy support and decision
De Vries, W., Reinds, G.J., Posch, M., 1994. Assessment of critical loads and their
making in the European Union. Ecosyst. Serv. 1, 31e39.
exceedance on European forests using a one-layer steady-state model. Water
Maes, J., Liquete, C., Teller, A., Erhard, M., Paracchini, M.L., Barredo, J.I., 2016. An
Air Soil Pollut. 72 (1e4), 357e394.
indicator framework for assessing ecosystem services in support of the EU
De Vries, W., 2014. Nutrients trigger carbon storage. Nat. Clim. Change 4, 425e427.
Biodiversity Strategy to 2020. Ecosyst. Serv. 17, 14e23.
De Vries, W., Du, E., Butterbach-Bahl, K., 2014. Short and long-term impacts of ni-
Masera, O.R., Garza-Caligaris, J.F., Kanninen, M., Karjalainen, T., Liski, J., Nabuurs, G.J.,
trogen deposition on carbon sequestration by forest ecosystems. Curr. Opin.
Pussinen, A., de Jong, B.H.J., Mohren, G.M.J., 2003. Modeling carbon seques-
Environ. Sustain. 9e10, 90e104.
tration in afforestation, agroforestry and forest management projects: the
De Wit, H.A., Autnes, K., Hylen, G., Dalsgaard, L., 2015. A carbon balance of Norway:
CO2FIX V.2 approach. Ecol. Model. 164 (2e3), 177e199, 15 June 2003.
terrestrial and aquatic carbon fluxes. Biochemistry 123, 147e173.
Moiseyev, A., Solber, B., Kallio, M.I., Lindner, M., 2011. An economic analysis of the
Dominici, F., Peng, R.D., Barr, C.D., Bell, M.L., 2010. Protecting human health from air
potential contribution of forest biomass to the EU RES target and its implica-
pollution: shifting from a single-pollutant to a multi-pollutant approach.
tions for the EU forest industries. J. For. Econ. 17 (2), 197e213.
Epidemiology 21 (2), 187e194, 2010 Mar.
National Research Council, 2005. Valuing Ecosystem Services: Toward Better
EC, 2009. Draft Commission Staff Document SEC(2009) 1710 Final Impact Assess-
Environmental Decision-Making. The National Academies Press, Washington,
ment Accompanying the Commission Decision Determining a List of Sectors
DC.
and Subsectors Which Are Deemed to Be Exposed to a Significant Risk of Car-
Nilsson, J., 1988. Critical Loads for Sulphur and Nitrogen. Swedish EPA).
bon Leakage Pursuant to Article 10a (13) of Directive 2003/87/EC C(2009)10251
Nordhaus, W., 2013. Estimates of the social cost of carbon: concepts and results
Final.
from the DICE-2013R modeland alternative approaches. J. Assoc. Environ.
EEA (European Environment Agency), 2015. Air Quality in Europe d 2015 Report.
Resour. Econom. 1 (No.1/2), 273e312.
EEA Report No 5/2015. EEA, Copenhagen, Denmark.
Papadimitriou, F., 2012. Artificial intelligence in modelling the complexity of
EPA, 1999. Costs and Benefits of the Clean Air Act. EPA, Washington DC.
mediterranean landscape transformations. Comput. Electron. Agric. 81, 87e96.
EPA, 2011. Benefits and Costs of the Clean Air Act 1990-2020, the Second Pro-
Piot-Lepetit, I., M'Barek, R., 2011. Methods to Analyse Agricultural Commodity Price
spective Study. EPA, Washington DC.
Volatility. Springer Verlag, New York.
Evans, C.D., Caporn, S.J.M., Carroll, J.A., Pilkington, M.G., Wilson, D.B., Ray, N.,
Posch, M., Aherne, J., JP Hettelingh, 2011. Nitrogen critical loads using biodiversity-
Cresswell, N., 2006. Modelling nitrogen saturation and carbon accumulation in
related critical limits. Environ. Pollut. 159, 2223e2227.
heath land soils under elevated nitrogen deposition. Environ. Pollut 143,
Reich, P., Hungate, B., Luo, Y., 2006. Carbon-nitrogen interactions in terrestrial
468e478.
ecosystems in response to rising atmospheric carbon dioxide. Annu. Rev. Ecol.
Fleischer, K., Wårlind, D., van der Molen, M.K., Rebel, K.T., Arneth, A., Erisman, J.W.,
Evol. Systemat. 37, 611e636.
Wassen, M.J., Smith, B., Gough, C.M., Margolis, H.A., Cescatti, A., Montagnani, L.,
Remme, R.P., Edens, B., Schro €ter, M., Hein, L., 2015. Monetary accounting of
Arain, A., Dolman, A.J., 2015. Low historical nitrogen deposition effect on carbon
ecosystem services: a test case for Limburg province, the Netherlands. Ecol.
sequestration in the boreal zone. J. Geophys. Res. BioSci. 120, 2542e2561.
Econ 112, 116e128.
 L. Hein et al. / Journal of Environmental Management 206 (2018) 650e663
Schro€ ter, M., Barton, D.N., Remme, R.P., Hein, L., 2014. Accounting for capacity and
flow of ecosystem services: a conceptual model and a case study for Telemark,
Norway. Ecol. Indic 36, 539e551.
Shaver, G.R., Chapin, F.S., 1995. Longeterm responses to factorial, NPK fertilizer
treatment by Alaskan wet and moist tundra sedge species. Ecography 18 (3),
259e275.
Simpson, D., Benedictow, A., Berge, H., Bergstrom, R., Emberson, L.D., Fagerli, H.,
Flechard, C.R., 2012. The EMEP MSC-W chemical transport model e technical
description. Atmos. Chem. Phys. 12, 7825e7865.
Skjelkvåle, B.L., Mannio, J., Wilander, A., Andersen, T., 2001. Recovery from acidifi-
cation of lakes in Finland, Norway and Sweden. Hydrol. Earth Syst. Sci. 5,
273e541.
Smart, J.C.R., Hicks, K., Morrissey, T., Heinemeyer, A., Sutton, M.A., Ashmore, M.,
2011. Applying the ecosystem service concept to air quality management in the
UK, a case study for ammonia. Environmetrics 22, 649e661.
Smith, W.H., 1990. Air Pollution and Forests: Interaction between Air Contaminants
and Forest Ecosystems, second ed. SpringerVerlag, New York, p. 618.
Spash, C.L., 2008. How much is that ecosystem in the Window? The one with the
bio-diverse trail. Environ. Val. 17, 259e284.
Statistics Norway, 2015. Website https://www.ssb.no/en/forskning/energi-og-
miljookonomi/baerekraftig-utvikling/experimental-urban-ecosystems-
accounting-urban-eea-improving-the-decision-support-relevance-for-
municipal-planning-and-policy, accessed 18 August 2017.
Stern, N., 2008. The economics of climate change. In: American Economic Review:
Papers & Proceedings, vol. 98, pp. 2e37, 2008.
Stevens, C.J., Thompson, K., Grime, J.P., Long, C.J., Gowing, D.J., 2010. Contribution of
acidification and eutrophication to declines in species richness of calcifuge
grasslands along a gradient of atmospheric nitrogen deposition. Funct. Ecol. 24
(2), 478e484.
663
Suwarno, A.L., Hein, Sumarga, E., 2016. Who benefits from ecosystem services? A
case study for Central Kalimantan, Indonesia. Environ. Manag. 57 (2), 331e344.
TEEB, 2010. The economics of ecosystems and biodiversity. In: Kumar, Pushpam
(Ed.), Earthscan. Ecological and Economic Foundations, London and Washing-
ton DC.
Thompson, T.M., Rausch, S., Saari, R.K., Selin, N.E., 2014. A systems approach to
evaluating the air quality co-benefits of US carbon policies. Nat. Clim. Change 4,
917e923.
Tromborg, Sjolie, 2011. Data Applied in the Forest Sector Models NorFor and NTMIII.
INA fagrapport 17. Department of Ecology and Natural Resource Management,
Norwegian University of Life Sciences, Aas, Norway.
UN, EC, FAO, OECD, 2014. World Bank. System of Environmental-economic Ac-
counting 2012, Experimental Ecosystem Accounting. United Nations, New York,
USA.
Van den Bergh, J.C.J.M., Botzen, W.J.W., 2014. A lower bound to the social cost of CO2
emissions. Nat. Clim. Change 4, 253e258.
Varian, H., 1992. Micro-economic Analysis. W. W. Norton & Company, p. 563.
Villamagna, A.M., Angermeier, P.L., Bennett, E.M., 2013. Capacity, pressure, demand,
and flow: a conceptual framework for analyzing ecosystem service provision
and delivery. Ecol. Complex. 15, 114e121.
Wamelink, G.W.W., Wieggers, H.J.J., Reinds, G.J., Kros, J., Mol-Dijkstra, J.P., van
Oijen, M., de Vries, W., 2009. Modelling impacts of changes in carbon dioxide
concentration, climate and nitrogen deposition on NPP and carbon sequestra-
tion of intensive forest monitoring plots in Europe. For. Ecol. Manag. 258,
1794e1805.
Zank, B., Bagstad, K.J., Voigt, B., Villa, F., 2016. Modeling the effects of urban
expansion on natural capital stocks and ecosystem service flows: a case study in
the Puget Sound, Washington, USA. Landsc. Urban Plan. 149, 31e42.