47 Open Environmental Sciences, 2008, 2, 47-53
1876-3251/08 2008 Bentham Open
The Effect of Changing Background Emissions on External Cost Estimates
for Secondary Particulates
L.L.R. Int Panis*
Integrated Environmental Studies, Flemish Institute for technological Research (VITO), Belgium
Abstract: This paper discusses the importance of background concentrations of NH3, SO2 and NOx for the estimation of
environmental external costs of secondary particulates. A modified version of the ECOSENSE software was developed
within the ongoing series of European ExternE projects, devoted to the assessment of energy related environmental exter-
nal costs. Using the Windrose Trajectory Model the yearly average concentrations of pollutants at ground level was calcu-
lated based on average meteo data and a simple scheme of atmospheric reactions. After this, epidemiological exposure-
response functions are applied to determine the impact on the receptors. Finally, the calculated physical impacts are
monetized on the basis of selected economic evaluations. The fact that estimates of external costs of incremental emis-
sions of NOx and SO2 will increase when background emissions decrease is the most important new result. The choice of
relevant background emissions is therefore essential to obtain meaningful estimates of external costs.
ants in the atmosphere are the sum of the costs (adverse im-
pacts) that are imposed on society and the environment but
not taken into account by the polluter (e.g. health effects of
inhalation of secondary aerosols). The European ExternE
project series is the most state-of-the-art attempt to build a
consistent methodology to estimate the environmental exter-
nal costs from different fuel cycles. The resulting accounting
framework was implemented in the ECOSENSE software
package. During several of the projects in the ExternE series,
different versions of ECOSENSE were distributed to partici-
pants in order to make calculations in their own countries
(E.g. the 1995 National Implementation project) . This
proved to be a very successful strategy leading to numerous
external cost estimates in different countries being published.
The software was distributed with databases including all
necessary data and parameters such as meteorology, back-
ground emissions, concentration response functions and
valuation data that allow the model to be used with default
settings. The parameters used for modelling however are
rarely discussed although this may have a large impact on
the results . The current European CASES project will
therefore need to take these considerations into account.
Environmental externalities from the emission of pollut-
SENSE is the Windrose Trajectory Model (WTM), based on
the windrose approach of the Harwell Trajectory Model. It is
used to estimate the concentration and deposition of secon-
dary acid species on a European wide scale. It was originally
The regional dispersion model at the heart of ECO-
*Address correspondence to this author at the Integrated Environmental
Studies, Flemish Institute for technological Research (VITO), Belgium;
developed at Harwell Laboratory by  for atmospheric ni-
trogen species, and extended to include sulphur species by
. The model is a receptor-orientated Lagrangian plume
model employing an air parcel with a constant mixing height
of 800 m moving with a representative wind speed. The re-
sults are obtained at each receptor point by considering the
arrival of 24 trajectories weighted by the frequency of the
wind in each 15° sector. The trajectory paths are assumed to
be along straight lines and are started at 96 hours from the
receptor point . All data required to run the Windrose Tra-
jectory Model, including the background emissions, are pro-
vided by the ECOSENSE database. The background emis-
sions are the total emissions from all other sources of SO2,
NOx and NH3 given on a 50x50km grid covering the whole
of Europe. These emissions undergo the same chemical
transformations as the pollutant source that is being studied
and determine the rate at which pollutants react in the at-
to resemble the atmospheric chemistry of the Harwell Tra-
jectory Model. The chemical transformations modelled in
WTM are schematically shown in Fig. (1). Through the ac-
tion of highly active oxygen containing species such as
ozone, primary emissions of NO and SO2 are converted into
acids. Reactions with ammonia transform these acids into
ammonium salts (ammonium nitrate and ammonium sul-
phate). Several chemical species occur in more than one dia-
gram in Fig. (1) (not all relationships are shown) indicating
competing reactions. Sulphuric acid is assumed to be some-
what more successful in scavenging ammonia from the at-
mosphere than nitric acid. All secondary species are removed
from the atmosphere through wet or dry deposition, each
species having its specific deposition rates.
Within ECOSENSE applications, the WTM is configured
not allow the modelling of organic aerosol formation. Never-
The simple reaction scheme used by ECOSENSE does
48 Open Environmental Sciences, 2008, Volume 2 L.L.R. Int Panis
theless  could show that (with the exception of organic
aerosols) WTM was sufficiently reliable given its purpose.
To test the validity of the annual average concentrations ob-
tained by ECOSENSE they were compared with measure-
ments as well as results of a more sophisticated model built
around the chemical gas phase mechanism CACM  and
the aerosol module MADRID 2 [8, 9] which provides un-
precedented accuracy for many components of the aerosol
mixture. Their comparison, complicated by the large number
of substances modelled by MADRID 2 and not by ECO-
SENSE , therefore focused on the total amount of secon-
dary inorganic pollutants and found that the results of both
models agreed well, especially for the fraction attributed
with health effects.
Health Effects of Secondary Aerosols
ogy and medical experiments, that particles in ambient air
cause adverse health effects in populations exposed to them
. A large number of studies have demonstrated an asso-
ciation between PM, respiratory and cardiovascular illnesses
and increased mortality (both acute and chronic). The most
important health endpoints used in ECOSENSE are summa-
rized in the first column of Table 1. ECOSENSE uses con-
It has been shown in epidemiology, but also in toxicol-
centration-response functions (CRF, references in Table 1)
and data on population density to estimate impacts on public
Table 1. Most Important Health Endpoints Included in Ex-
ternE (Summarized from )
Health Endpoint Unit Implicated Pollutants
YOLL5 Primary PM2.5 (other)
Primary PM2.5 (traffic)1
Nitrate aerosols 3
Sulphate aerosols 4
Cases Primary PM2.5
No effect included Nitric Acid
Sec. Organic aerosols
1Assumed to cause 50% more mortality than the general PM2.5 mixture.
2Assumed to cause 40% less mortality than PM2.5.
3Assumed to cause 50% less mortality than PM10.
4Assumed to be equivalent to PM10 (including sulphuric acid).
5Years Of Life Lost (loss of life expectancy).
Fig. (1). Schematic view of atmospheric reactions in ECOSENSE describing the transformation of nitrogen and sulphur containing emissions
into secondary aerosols (source: adapted from  by ).
External Cost Estimates for Secondary Particulates Open Environmental Sciences, 2008, Volume 2 49
on the mass of PM without any distinction of physical or
chemical characteristics (acidity, solubility, …). There is for
example limited evidence for the toxicity of components like
sulphates and nitrates that make up a large fraction of the
mass of particulate matter in ambient air. In particular there
is a lack of epidemiological studies related to nitrate aerosols
because until recently this pollutant was not monitored by air
pollution monitoring stations . It is however crucial to
quantify the external costs of source-specific contributions to
ambient PM. This will improve decisions on cost-efficient
abatement strategies, and will allocate the scarce resources to
those emission reduction measures that yield the largest re-
duction in air pollution related health effects.
Most of the available epidemiological studies are based
and secondary particles. In a first series of studies  the
assumption was made that the toxicity of all sulphates was
equal to that of the general PM2.5 mixture and the toxicity of
particulate nitrates equal to that of PM10. This distinction
between sulphates and nitrates was based only on size, not-
ing that nitrates need other particles to condense on, whereas
sulphates self-nucleate and are therefore smaller on average.
The ratio of CRF slopes of PM10 to PM2.5 was taken as 0.6,
because this is a typical value of the ratio of concentrations
of PM2.5 and PM10 .
Therefore ExternE tried to differentiate between primary
ternE series the assumptions about the toxicity of the differ-
ent PM types have been adapted to the latest epidemiological
and toxicological evidence.  state that it is now more
likely than not that primary particles from combustion, more
specifically from traffic are more damaging to health than
other particles. For the secondary particles the evidence is
less convincing. In particular for nitrates there is still not
much evidence for harmful effects, whereas for sulphates
quite a few studies, including the very important cohort
study of , do find associations. Thus sulphates are now
treated as PM10, and nitrates as being half as toxic as PM10.
Primary particles from traffic are estimated to be 1.5 as toxic
as PM2.5 and particles from other combustion sources like
power plants are also treated as PM10.
For the most recent methodological report from the Ex-
Implications for External Cost Estimates in the Future
may yield results that are not compatible with the most up-
to-date methodology. Ever since the first model versions
were distributed, important changes to the ExternE method-
ology have been made. Several dominant exposure-response
functions (e.g. chronic mortality and chronic bronchitis)
were revised a number of times and recently new empirical
results for the value of a Life Year lost were introduced (e.g.
[5, 11, 14-16]. While the exposure-response functions and
monetary values included with the ECOSENSE software can
easily be modified by the user, changing the background
emissions used by the Windrose Trajectory Model is not
straightforward. More recent background emissions have
been compiled into ECOSENSE compatible formats, but
these were not widely distributed among ECOSENSE users
from previous ExternE projects (Bickel P., pers. com..).
Most users therefore still use the default background emis-
The continued use of older ECOSENSE versions today
sions data included with the model that refer to the years
1990 or 1994. Even the ECOSENSE Pointsource 2000 ver-
sion still contained 1994 background emissions albeit on a
50x50km resolution (Krewitt W., pers. comm.).
1994 has meanwhile been updated by EMEP. The first ver-
sion will therefore be referred to as 1994a (included with
ECOSENSE) and the more recent update (used in this paper
for the first time) as 1994b.
To complicate things, the emissions database for the year
MATERIALS AND METHODS
obtained with the transport version of ECOSENSE .
These results also apply to the point source versions because
the WTM parameters appear to be the same in all versions
available to the author. In addition a preliminary assessment
of the deposition and wash-out modelling in both versions
revealed no meaningful differences . The ECOSENSE
software is available from the Institut für Energiewirtschaft
und Rationelle Energieanwendung (IER) of the University of
Stuttgart in Germany. More details can be found at: http:
The modeling results discussed in this paper were all
on the EUROGRID co-ordinate system, which defines equal-
area projection grid cells, covering all EU and European
non-EU countries in a Paradox database at a 50x50km reso-
lution . All versions known to us use either 1990 or
1994a EMEP emissions.
ECOSENSE stores only one set of background emissions
for the years 1990, 1994b, 2000, 2005, 2010 and 2020 were
taken for the EU18 countries (including the 15 old member
states as well as Iceland, Norway and Liechtenstein) from
the UNECE/EMEP emission database WebDab (http:
//www.emep.int). This online database was constructed to
facilitate access to the emission data reported to the Conven-
tion on Long-Range Transboundary
(CLRTAP). This data served as the basis for the preparation
of six new versions of the background emissions Paradox
database that were integrated to work with ECOSENSE and
WTM. This involved running the model without any extra
emissions to create a database with background concentra-
tions which was then again integrated to work with ECO-
SENSE and WTM. Results were calculated for each dataset
and used in a validation of previously published results.
For the calculations presented here, new emissions data
Total Background Emissions (EU18)
erably since 1990 (Fig. 2) and this is expected to continue for
another decade. The total decrease between 1990 and 2020 is
projected to be 80% for SO2 and 60% for NOx. For individ-
ual grid cells (e.g. in Belgium) the differences may be even
Emissions of SO2 and NO2 have been decreasing consid-
extent. Projected emissions of NH3 were not completely re-
ported by all countries for the year 2005 and could therefore
not be used in this study. This different evolution will
change balance in the atmospheric chemistry and change the
The emissions of NH3 have decreased to a much lesser
50 Open Environmental Sciences, 2008, Volume 2 L.L.R. Int Panis
amount of SO2 and NOx that gets converted into secondary
particulate matter . This effect is reproduced in a fairly
simple way by the WTM reaction scheme (Fig. 1).
Externalities Resulting from Secondary Aerosols
the release of 1 additional tonne of pollutant. Six runs with
different background emissions were performed. The results
are expressed in Euro per tonne of pollutant emitted (either
NOx or SO2) and includes all damages caused by the secon-
dary aerosols resulting from this primary emission.
Fig. (3) shows the results obtained with ECOSENSE for
background emissions originally provided with ECOSENSE)
are quite similar to the results obtained with the emissions
for 1994 (1994b) currently distributed by EMEP. This exer-
cise therefore constitutes an important independent valida-
tion of the ExternE-methodology. Fig. (3) also shows that
costs per tonne are expected to rise significantly in the fu-
ture. This effect can easily be understood from the ECO-
SENSE reaction scheme (Fig. 1). Because the emissions of
NH3 are expected to stay at a similar level whereas SO2
emissions show a strong decrease (Fig. 2) more NH3 is left to
react with sulphuric acid and transform it to ammonium sul-
The original results (labelled 1994a, obtained with the
more NH3 is left to react with nitric acid and more nitric acid
will be converted to nitrate aerosol. Since ECOSENSE only
attributes health effects to the nitrate aerosol (and not to its
nitric acid precursor) the cost per one tonne of NOx emitted
Similarly when NOx emissions decrease significantly
2020 than it was in 1990. The emission of one tonne of SO2
will also lead to impacts from sulphates that are nearly twice
as high in 2020 as compared to 1990.
Model results show it to be more than twice as high in
with ECOSENSE was done in the Green Accounting project
[5, 19] where it was demonstrated that impacts from emis-
sions are higher when emissions in a neighbouring country
are set to zero (a test was carried out for Germany and the
Netherlands). The reason for this effect is that concentrations
of NH3 are higher without the emissions in the neighbouring
country so that more NH3 remains available to react with
local SO2 and NOx emissions to form sulphates and nitrates.
The influence of German emissions on the Dutch damage
estimation was about 10% and vice versa about 1 percent. In
addition  demonstrated that the effects of emission re-
ductions on impacts are not linear, which is entirely consis-
tent with the results presented here (Fig. 3).
The only similar manipulation of background emissions
external cost estimates (in ?/tonne of primary pollutant emit-
ted) will rise in the future. This is due to two mechanisms in
the scheme of atmospheric reactions used in the WTM mod-
ule. The first effect is that more sulphuric acid is converted
to ammonium sulphate when more ammonia is available.
This in turn causes more SO2 to be converted to sulphuric
acid. Both sulphuric acid and ammonium sulphates are at-
tributed with adverse health effects in ECOSENSE. These
health impacts from the release of 1 unit of SO2 will increase
54% between 1994(b) and 2010. On the other hand impacts
from SO2 itself will be 10% lower because it gets depleted
from the atmosphere. The second effect is that similarly
more nitric acid is converted to ammonium nitrate when NH3
is abundant. In this way future concentrations of nitric acid
will be lower and concentrations of ammonium nitrate will
rise. This finding is related to the question whether nitric
It is clear from the results in Fig. (3) that ECOSENSE
Fig. (2). Comparison of total emissions over the EMEP grid for different years.
External Cost Estimates for Secondary Particulates Open Environmental Sciences, 2008, Volume 2 51
acid is also toxic (pers. comm. A. Rabl, 2004). Unlike sul-
phuric acid, the tested ECOSENSE version assumes nitric
acid is harmless (P. Bickel, pers. comm., 2004). This ex-
plains why impacts from this pathway are expected to show
a stronger increase (+87% between 1994(b) and 2010). If
however nitric acid is toxic, the present day impacts have
been underestimated and the extra conversion of nitric acid
to nitrate in the future would not have an additional effect. If
nitric acid is toxic, the damage cost attributed to the emission
of NO2 could be about 15 to 35% higher than with the refer-
ence assumption of ExternE. The reason why the difference
is not larger lies in the high deposition velocity of HNO3
which reduces the geographic range of the impact relative to
particulate nitrates P. Bickel (pers. comm., 2004).
compared with the values given in , Table 1 for Belgium
(1990 – 2010). The change they report for nitrates is very
similar (93% versus 96% increase in this calculation). The
increase in cost per tonne of SO2 through exposure to sul-
phates was only predicted to increase by 6%-18% by ,
while it was seen to increase 87% in this trial. The most
likely explanation for this difference is that different back-
ground concentrations were used as emissions predictions
for 2010 were updated by EMEP. It proved however impos-
sible to find out exactly which data had been used and com-
pare it with our approach. The continuous update of back-
ground emission files and the use of the correct files for the
questions under study should therefore be a constant point of
attention. Authors should always state which background
emissions were used and be aware of the consequences of
using outdated background emissions for their results. The
issue of changing background emissions is probably even
more important for ozone formation and its impacts (see 
for a discussion).
The relative changes found in our calculations were
Implications for the Results Published in the ExternE
widely been used by policy makers to decide on the best
automotive technologies (from an environmental point of
view) and to estimate total savings from different policy
proposals (by linear aggregation of incremental external-
ities). Implicitly external costs per tonne of pollutant calcu-
lated for a given year (often 1990 or 1994) were used to es-
timate avoided externalities in the future. Nevertheless it is
important to take into account changes in atmospheric condi-
tions when analyzing policy decisions that take their full
effect in 2010 or later. Using external cost data obtained with
software using 1990 or 1994 background emissions could
yield spurious results, depending on which pollutants (pri-
mary or secondary) cause the dominant impacts.
Results from the ExternE transport project  have
tical problems, we have recalculated the external costs for a
diesel car driven in the centre of Brussels (Belgium), one of
the case studies explored in detail in . The results shown
in Fig. (4) (expressed as ?-cents per vehicle.km) were ob-
tained by multiplication of the cost per tonne and specific
emission factors. Only results for secondary pollutants are
shown. Direct health impacts from SO2 and impacts on soil
are so small that they are indiscernible on this scale. It is
clear that for the analysis of transportation problems the
changing value of sulphate impacts can be neglected when
compared to increasing nitrate impacts. The sulphur content
of all currently available transportation fuels has become so
low under European fuel standards that the impacts have
become very small. Impacts attributed to NOx emissions
could in this case be approximately 0,3 ?-cent higher than
previously reported. Obviously the total cost per km is less
sensitive to the non-linearity of the secondary aerosol forma-
tion. Depending on the magnitude of impacts resulting from
To demonstrate the relevance of our main result to prac-
Fig. (3). Damage cost per tonne emitted under different background conditions (years).
52 Open Environmental Sciences, 2008, Volume 2 L.L.R. Int Panis
primary emissions this may or may not be an important dif-
ference. For the majority of the results published in ExternE
transport  primary PM emissions accounted for the largest
share of the total impacts by far, and changing the value for
nitrate would not have a significant effect on the total costs.
filters in advanced and future vehicles, the comparison of
technologies and modes for future years often depends on a
correct estimate for the external effects of NOx .
On the other hand with the increased use of particulate
Implications for Industrial Emissions from High Stacks
source model have been widely used to compare different
technologies for electricity production. European energy
policy has partly been influenced by results from the Ex-
ternE projects. We believe that the ECOSENSE point source
model behaves very similar to what was demonstrated in this
paper for transport related problems (see also ). Never-
theless there are some important differences. Emissions from
power plants are mostly emitted from high stacks that are
some distance away from populated areas whereas exhaust
gases from traffic are sometimes contained within street can-
yons before being dispersed. The sulphur content of the fuels
used is also considerably higher than that of present day
transportation fuels. The external costs of power plants are
therefore dominated by impacts from secondary pollutants.
The correct use of relevant background emissions with
ECOSENSE point source is therefore even more important
for applications focusing on power plants.
As for transport, results from the ECOSENSE point
external costs caused by ammonium sulphate and ammo-
nium nitrate in the future (formed from the emission of NOx)
The most important new finding in this paper is that the
are underestimated when using wrong background emissions
with ECOSENSE or any other model.
effect depends on how nitric acid is dealt with in the epide-
miological calculations. If nitric acid is considered not to be
toxic, incremental health impacts of secondary nitrogen spe-
cies will more than double between 1990 and 2010. If on the
other hand nitric acid is attributed with adverse health ef-
fects, present day impacts are underestimated, but the ex-
pected relative increase in the future will be smaller. Exter-
nal costs from SO2 emissions (through the adverse effects of
ammonium sulphate on public health) will increase by 87%
between 1994 and 2010.
There is some uncertainty because the magnitude of this
pollutants, results generated for present transport technolo-
gies show little change.
Given the dominance of PM2.5 and other locally acting
for calculations related to advanced vehicles such as gaso-
line-electric hybrids and diesels with particulate filters.
However, the results presented here are highly relevant
ants for point source emissions from high stacks, changes to
the externalities from power plants are expected to be impor-
Because of the relative importance of secondary pollut-
used to produce more accurate estimates of externalities
within the ongoing ExternE projects NEEDS and CASES
The findings presented in this paper will therefore be
to the results presented in this paper.
Discussions with Peter Bickel and Ari Rabl contributed
Fig. (4). Results for the regionally acting pollutants (from Belgian case study of a 10km trajectory through Brussels, ExternE 1998 .
External Cost Estimates for Secondary Particulates Open Environmental Sciences, 2008, Volume 2 53
Nocker, Daan Beheydt, Carolien Beckx and four anonymous
reviewers for their useful comments on earlier versions of
this paper. This work was partly funded under the European
ExternE-POL project in the ExternE series.
The author also wishes to thank Rudi Torfs, Leo De
 ExternE. Externalities of Energy. National Implementation. Brus-
sels: European Commission EC DGXII, Science Research and De-
velopment, 1999; EUR 18528: 606 pp. 606.
Brizio E, Genon G. The influence of different mixing heights on
the ECOSENSE model results at a local scale. Environ Model
Software 2005; 20(7): 917-33.
Derwent R, Nodop K. Long-range transport and deposition of
acidic nitrogen species in north-west Europe. Nature 1986; 324:
Derwent R, Dollard G, Metcalfe S. On the nitrogen budget for the
United Kingdom and north-west Europe. QJR Meteorol Soc 1988;
Friedrich R, Bickel P, Ed. Environmental External Costs of Trans-
port. Heidelberg: Springer Verlag, 2001.
Int Panis L, Deutsch F, Torfs R. Factors contributing to the uncer-
tainty of secondary particulates externalities: 2006: Proceedings of
the 2nd International Conference on Quantified Eco-Efficiency
Analysis for Sustainability: 2006; Egmond aan Zee, The Nether-
Griffin R, Dabdub D, Seinfeld J. Secondary organic aerosol 1.
Atmospheric chemical mechanism for production of molecular
constituents. J Geophys Res 2002; 107(D17): 4332.
Pun B, Griffin R, Seigneur C, Seinfeld J. Secondary organic aero-
sol 2. Thermodynamic model for gas/particle partitioning of mo-
lecular constituents. J Geophys Res 2002; 107(D17): 4333.
Zhang Y, Pun B, Vijayaraghavan K, et al. Development and appli-
cation of the Model of Aerosol Dynamics, Reaction, Ionization,
and Dissolution (MADRID). J Geophys Res 2004; 109: D01202.
 WHO. WHO air quality guidelines global update: WHO Regional
Office for Europe, Denmark, 2005.
Rabl A, Hurley F. Impact Pathway approach: Exposure Response
functions. In: Bickel P, Friedrich R, Ed. ExternE Externalities of
Energy, Methodology 2005 Update. Brussels: European Commis-
sion, DG Research 2005; 270.
Pope C, Burnett R, Thun M, et al. Lung cancer, cardiopulmonary
mortality and long-term exposure to fine particulate air pollution. J
Amer Med Assoc 2002; 287(9): 1132-41.
Abbey D, Lebowitz M, Mills P, Petersen F, Lawrence Beeson W,
Burchette R. Long-term ambient concentrations of particulates and
oxidants and development of chronic disease in a cohort of non
smoking California residents. Inhal Toxicol 1995; 7: 19-34.
ExternE. Methodology Report, 2nd Edition Volume 7. Brussels:
European Commission EC DGXII. Sci Res Develop 1998.
Int Panis L, De Nocker L. Marginal Costs Belgium. In: Friedrich R,
Bickel P, Ed. Environmental External Costs of Transport. Heidel-
berg: Springer Verlag 2001; 169-73.
Int Panis L. Impacts on crops. In: Bickel P, Friedrich R, Ed. Ex-
ternE Externalities of Energy, Methodology 2005 Update. Brussels:
European Commission. DG Res 2005: 270.
Bonnefous S, Despres A. Evolution of the European data base:
IPSN/EURATOM - CEA Association, BP 6, 92265 Fontenay-Aux-
Roses, France, 1989.
Erisman J, Schaap M. The need for ammonia abatement with re-
spect to secondary PM reductions in Europe. Envir Pollut. 2004;
GARPII. Green Accounting In Europe, A Comparative Study.
Edgar Elgar, Cheltenham: The Fondazione Eni Enrico Mattei (FE-
EM); 2005; Vol 2.
Krewitt W, Trukenmuller A, Bachman TM, Heck T. Country-
specific damage factors for air pollutants. Int J LCA 2001; 6(4):
Int Panis L, De Nocker L, De Vlieger I, Torfs R. Trends and uncer-
tainty in air pollution impacts and external costs of Belgian passen-
ger car traffic. Int J Vehicle Des 2001; 27(1-4): 183-94.
Received: February 25, 2008
Revised: April 7, 2008 Accepted: April 11, 2008
© L.L.R. Int Panis; Licensee Bentham Open.
This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/license/by/2.5/), which
permits unrestrictive use, distribution, and reproduction in any medium, provided the original work is properly cited.