PAPER www.rsc.org/jem | Journal of Environmental Monitoring
Water quality indices across Europe a comparison of the good
ecological status of five river basinsw
¨ ¨
Peter Carsten von der Ohe,*a Andrea Pruß,ab Ralf Bernhard Schafer,cd
Matthias Liess,c Eric de Deckeree and Werner Bracka
Received 28th March 2007, Accepted 6th July 2007
First published as an Advance Article on the web 30th July 2007
DOI: 10.1039/b704699p
The European Water Framework Directive (WFD) requires the definition of near-natural
reference conditions to determine the extent of water bodies deviation from   good ecological
status  caused by stress gradients. However, the classification of ecological quality depends on the
assessment method applied and the stressor concerned. While assessment methods that are
generally applicable would be favourable, many European countries employ the locally developed
water quality metrics that assess the impact of organic pollution (including eutrophication) and
the associated decrease in dissolved oxygen. These indices do not specifically address stress from
organic toxicants, such as pesticides. The aim of this study was to examine the performance of
presently used assessment methods to identify reference conditions of non-contaminated streams
in five selected European river basins, covering the geographical region from Spain to Finland, as
a crucial prerequisite to indicate toxic gradients. The analysis comprised the Belgium biotic index
(BBI), the biological monitoring working party (BMWP) scoring system and the revised German
saprobic index. For comparison, we included an adaptation of the recently developed SPEAR
index. In two previous field studies, this metric highly correlated with measured pesticide
gradients. In this study, SPEAR was the only indicator that was generally applicable to all
monitoring data and capable of determining   high ecological status  of reference conditions in all
basins. Thus, based upon previous and own results, the authors suggest the species at risk
(SPEAR) index to be potentially useful as a European-wide index to address deviations from
  good ecological status  due to organic toxicants and recommend it for consideration in
integrated water-resource evaluations under the WFD.
true measure of the impact, they should indicate a degradation
Introduction
gradient according to a dominant stressor.2,3 A metric that
The protection and preservation of freshwater resources pose
should be applied at a large spatial scale like Europe, must
a major challenge to modern societies. The European Com-
therefore be robust to natural and spatial variability compared
mission acted in response and introduced the Water Frame-
to the change in the index values caused by the indicated
work Directive (WFD) as an instrument to sustain and
stressor.4,5
improve water quality.1 By 2015, the directive aims to achieve
In this context, stream-dwelling invertebrate communities
at least a   good ecological status  for all water bodies, i.e.
have a long history of being used as biological indicators to
streams that are   only slightly  impacted by anthropogenic
assess the quality of surface waters6,7 and still represent the
stressors. Hence, human activities that result in aberration
most common used organism group.8,9 Consequently,
from this status have to be identified to implement effective
monitoring of resident macroinvertebrate communities has
programmes of measures.1 This, however, requires appropri-
become a primary component of water-resource evaluations
ate indicators that link observed effects to certain anthropo-
with regard to the WFD.1 A tendency to develop its own
genic stressors.2 Even though such indicators do not reflect a
national assessment methods could be observed for many
Member States.10 The Directive enables the Member States
a
Department of Effect-Directed Analysis, UFZ Helmholtz Centre for
to maintain their own methods, but outlines an intercalibra-
Environmental Research, Leipzig, Germany
b
tion procedure of the methods outputs, namely the classifica-
Department of Water Management, University of Applied Sciences
Magdeburg-Stendal, Magdeburg, Germany tion of   high  and   good ecological status  in the context of a
c
Department of System Ecotoxicology, UFZ Helmholtz Centre for
common implantation strategy.11 As one example, Birk and
Environmental Research, Leipzig, Germany
d Hering9 directly compared different assessment methods that
Institute for Ecology and Environmental Chemistry, University
are in current usage or those that are about being implemen-
Lüneburg, Lüneburg, Germany
e
Department Biology Ecosystem Management Research Group,
ted. They found significant intercorrelation for several metrics
University of Antwerp, Antwerp, Belgium
that lessens the added value of their simultaneous application.
w Presented at the Water Status Monitoring of Aquatic Ecosystems in
Nevertheless, it is generally accepted that multimetric assess-
the context of the Water Framework Directive meeting, Lille, France
12 14th March, 2007. ment methods provide better insight in the relationship
c
970 | J. Environ. Monit., 2007, 9, 970 978 This journal is The Royal Society of Chemistry 2007
between dominant stressors and ecological status.12,2 An index was highly correlated with the measured pesticide
example is the German Assessment System Macrozoobenthos gradients.3,20 The study of Scha¨ fer and colleagues20 included
that assesses ecological status of streams in three modules with two field surveys in the Scorff (France) and the Porvoonjoki
respect to the biological quality element (BQE) of benthic (Finland) River basins that were included in our analysis to
macroinvertebrates constrained by surface water body stream extend the geographical range.
type. With regard to effects of   organic pollution  , defined as
an increase in both organic components (e.g. increased biolo-
Materials and methods
gical oxygen demand (BOD); organic pollution sensu stricto)
and nutrient contents (eutrophication) by Sandin and
Studied river basins
Hering,13 the saprobic index is employed. Other countries like
The investigated streams covered different European eco-
Belgium or Spain apply single assessment methods to imple-
regions according to Illies,21 that spanned the geographical
ment the WFD. All these indices are especially powerful in
range from Spain to Finland. It comprised the Spanish
detecting the effects of oxygen depletion resulting from organic
Llobregat (ecoregion no. 1 Iberic-Macronesian region), the
pollution,13 but it is unclear if these indices are capable of
French Scorff (ecoregion no. 13 western plains), the Belgium
detecting adverse effects of organic toxicants. In a study of
part of the Scheldt (ecoregion no. 13), the German Weser
Schriever et al.,14 the saprobic index did not correlate with a
(ecoregions no. 9 central highlands and no. 14 central plains)
gradient of modelled pesticide input.
and the Finish Porvoonjoki (ecoregion no. 22 Fenno-Scan-
In this context, several supportive European Commission
dian shield).
research projects of the fifth and sixth Framework Programme
The Llobregat River represented a typical Mediterranean
address the impact of environmental pollutants, i.e. AQUA-
river basin, whereas the Scheldt River embodied a highly
TERRA, HAIR, MODELKEY, REBECCA or STAR. As
polluted northwest European river basin. Datasets were ob-
one example, the integrated project MODELKEY (511237-
tained from the Agencia Catalana de l Aigua (ACA, Barcelo-
GOCE) aims to develop models for assessing and forecasting
na, Spain) and the Vlaamse Milieumaatschappij (VMM,
the impact of environmental key pollutants on marine and
Erembodegem, Belgium), respectively. Moreover, the analysis
freshwater ecosystems,15 considering among other BQEs the
comprised the Weser River, representing a central European
group of benthic macroinvertebrates. To achieve this aim, the
river basin in the northern German lowlands that was pro-
available monitoring data from three case study river basins
vided by the Niedersa¨ chsischer Landesbetrieb fur Wasser-
¨
were collated and implemented in the central MODELKEY
wirtschaft, Kusten- und Naturschutz (NLWK, Hildesheim,
¨
database that was used in the present analysis.
Germany) and two field studies.20 From the latter two, the
According to the WFD, the assessment of the ecological
Scorff catchment represents a western European river basin
status of a water body undergoing monitoring requires a
that is located in the northwest of France and the Porvoonjoki
comparison of observed metric values to expected values of
catchment a north European river basin situated in the south
a predefined   reference condition  .1,16 In the context of the
of Finland.
WFD, the reference condition is termed   high ecological
status  and could be defined as group of sites being represen-
Selection of non-contaminated streams
tative of totally or nearly undisturbed streams.17,18 The iden-
tification of reference conditions is a crucial prerequisite to All selected sites were chosen according to criteria to reference
calculate ecological quality ratios and to determine the dele- sites of the STAR protocols regarding macroinvertebrates,2
terious effects of anthropogenic stress.19 with a focus on high chemical status. The mandatory criteria
As a first step to identify an appropriate indicator for toxic included (1) only minimal anthropogenic disturbances,
effects of organic compounds, the aims of this study were to (2) coarse woody debris should not be removed, (3) stream
examine (i) the value of currently employed assessment meth- bottoms and margins must not be fixed, (4) natural riparian
ods to correctly determine reference conditions of streams not vegetation, (5) no point sources of pollution, nutrient input or
contaminated by organic toxicants and (ii) the quality metrics eutrophication affecting the site and (6) no sign of acidification
general applicability to data of national river monitoring or salinity.
programmes from different European river basins. Finally, Selection was based upon information from topographical
(iii) the different metrics were analyzed for correlations to maps (1 : 10 000 1 : 50 000), i.e. existence of treatment plants,
avoid redundant indications. The analysis comprised the point sources of pollution or nutrients and other land use
Belgium biotic index (BBI) as the official standard method patterns (e.g. mining) led to the exclusion of sites. In the case
for biological water quality assessment as employed in the of the two field studies, the reference sites were selected
Flemish part of the Scheldt River, as well as the scoring system accordingly, in close cooperation with local authorities and
of the biological monitoring working party (BMWP) and its by inspection of sites (INRA, Rennes, France and SYKE,
adaptation for application to the Catalan Llobregat River, Helsinki, Finland). The above mentioned requirements
referred to as BMWP (SP). Furthermore, the revised German applied only to remote headwater streams and smaller tribu-
saprobic index, further referred to as SI (DE), was applied for taries located in forested catchments. These sites were con-
the Lower-Saxony part of the Weser River. An adaptation of sidered as being representative of non-contaminated streams
the recently developed species at risk (SPEAR) index,3 to of presumed high ecological status and are referred to as
detect effects of organic toxicants, was applied for comparison. reference sites in the following discussion. The selection of at
In two previous field studies (20 and 29 sites), the SPEAR least five of these high quality reaches per basin was hindered
c
This journal is The Royal Society of Chemistry 2007 J. Environ. Monit., 2007, 9, 970 978 | 971
by the limited availability of sampling data of the national project (EVK1-CT1999-00027). Respective metric values were
river monitoring programmes as well as own investigations.20 not transformed prior to the analyses.
It could therefore not consider specific biological reference
Species at risk of being affected by organic toxicants
conditions of all prevailing stream types in the five ecoregions.
A specific stream typology that was only available for German
In comparison to the other quality metrics, the recently
streams was used for an analysis of stream type dependency of
developed index of   species at risk  3 was originally developed
the metric results. Hence, the Weser reference sites were
to detect the effects of pesticides impacting aquatic commu-
assigned to stream types based upon the map Karte der
nities in the season of late April to June, the main application
biozonotisch bedeutsamen Fließgewassertypen Deutschlands
¨ ¨
period in the investigated area. The index was especially
(Stand Dezember 2003).
powerful in small streams with an agricultural catchment.
For the selection of a monitoring site as a reference site, at
The invertebrate species were classified at risk of being affected
least two recent macroinvertebrate samples were required.
by pesticides according to their physiological sensitivity to-
Samples were taken in accordance with local protocols which
wards organic toxicants31 as well as additional life history
were presumed to consider potential seasonality in the respec-
information,3 i.e. the recovery potential of species.32 For many
tive metric values (see below). Prior to analysis, the average
European species, this information is freely available from the
metric value for all samples was calculated at each site in order
internet (http://www.ufz.de/index.php?en=2138).
to avoid temporal pseudo-replication. For the Llobregat
However, macroinvertebrate communities in larger rivers
River, a total of 34 monitoring sites with samplings between
and streams with their numerous point and diffuse sources of
2001 and 2004 were available, five of which were classified as
organic environmental pollutants (e.g. effluents of chemical
reference sites. The 28 samples were taken in spring and
plants or intense agricultural areas) are constantly exposed to
autumn, before and after summer droughts, to account for
various compounds over the years. Therefore, in contrast to a
the arid climate. In this context, however, Zamora-Munoz
Ü
previous study,3 only the following two criteria were applied to
et al.22 stated no seasonality in the respective BMWP metric
classify species to be at risk of organic toxicants: a taxons S-
values. In the Weser River basin, from 279 national monitor-
value, representing its physiological sensitivity, greater than
ing sites with samplings between 1994 and 1999, eleven sites
 0.36 (median sensitivity of all species) and the generation time
with 38 samples were classified as reference sites. Please note
is equal or longer than half a year. Hence, the emergence of
that for this study, only samples for the months February to
insects was not taken into account. Note that this led to only a
August have been considered, as required by the AQEM
minor adaptation of the original SPEAR classification.3
method.12 Accordingly, seven of the 236 monitoring sites in
Limnephilus lunatus and Anabolia nervosa, like the whole
the Scheldt River with 36 sampling data from 2000 to 2004 met
family of Limnephilidae, were classified as being at risk. This
the above-mentioned requirements for a reference site.
holds also for the family Asselidae, due to the likely presence
Streams in the basin of the river Porvoonjoki and of the Scorff
of Asellus aquaticus as representative of this family. The low
River basin were each sampled in two successive months in
generation times and high reproduction rates of many Baetis
2005, and five and six sites were selected as reference sites,
and Gammarus species led to the classification of SPEnotAR of
respectively.
these genera. The same applies for the corresponding families
of Baetidae and Gammaridae. SPEAR was computed as a
ratio of the total abundance of species at risk and the total
Local indices of organic pollution
abundance of all species (eqn (1)):
In case of the Llobregat River, the locally adapted BMWP
Pn
(SP)23 was applied, whereas the original BMWP,24 reflecting
ai ti
central European conditions, was applied for the other basins.
SPEARźiź1 ð1Þ
Pn
Samples were collected from the riffle and pool habitats using
ai
an adaptation of the kick-net method and results obtained
iź1
were qualitative, on the taxonomic level of the family. The
where ai is the taxons abundance and ti is 1 for species
Belgian BBI25,26 required quantitative results at the genus
classified at risk, else 0.
level, expressed as number of individuals per dredge using a
To reduce the weights of highly abundant species, the
Van Veen dredge engine. The German SI (DE)27,28 was
originally reported abundance was transformed into seven
applied to historical monitoring data that were sampled
abundance classes in accordance with the conversion table of
according to the DIN 38410-1 normative with semi-quantita-
the revised German saprobic index.12 If only qualitative
tive results at the species level. The application of the revised
abundance information was available, as in the case of the
saprobic index led to a higher number of valid   historic 
monitoring data from the Llobregat River (presence of
samples, due to the higher number of classified taxa compared
families), an abundance class of one was assigned to each
to the old saprobic index.29 Species abundance was classified
family. For convenience, the calculation of the SPEAR index
into seven abundance classes.
was implemented into the MODELKEY database.
For consistency, the macroinvertebrate data of all basins
were taxonomically adjusted according to the AQEM guide-
Ecological status class boundaries
lines.30 The three metrics of organic pollution were automati-
cally calculated using the AQEM assessment software version For the classification of the ecological status of the investigated
3.0 (http//www.aqem.de) that was developed in the AQEM surface water bodies, the predefined class boundaries of the
c
972 | J. Environ. Monit., 2007, 9, 970 978 This journal is The Royal Society of Chemistry 2007
respective indices were taken from the legal guidance docu- anthropogenic stressors.19 Consequently, employing similar
ment.33 For the SI (DE), applied in Germany, stream type class sizes of 14%, the class boundaries were suggested as
specific class boundaries existed with regard to the river typol- r1%, 1.1 15, 15.1 29, and 29.1 43, corresponding to bad,
ogy.27 The selected reference conditions of the Weser basin poor, moderate, and good ecological status, respectively. The
consisted of sites of stream type no. 7 (small streams in lower low class boundary value of r1% may surprise, but it con-
mountainous areas in Central Europe) as well as stream types no. sidered the existence of potentially diverse communities at
14 (small sized, sand bottom streams in the German lowlands) disturbed sites with almost no species classified as being at risk
and no. 16 (small sized gravel bottom streams in the German of toxicants. The class boundaries suggested here, however,
lowlands). Predefined class boundary values for the SI (DE) are should be regarded as preliminary.
reported as 2.1, 2.25 and 2.15 for good ecological status as well as
1.6, 1.8 and 1.65 for high ecological status, respectively.34 The
Statistical analysis
predefined class boundaries for the good and the high ecological
status of the BBI were set to 7 and 9, respectively. The scoring
The   applicability  score of an index referred to the number of
system of BMWPs class boundary for good ecological status
countries where the index could be applied. Note, however,
correspond to values of 60 for the BMWP(SP) applied in the
that invertebrates of our own investigations in the Scorff River
Llobregat and 70 for the BMWP applied in the other basins,
basin were identified to a lower (genus) taxonomic level
respectively. Sites with BMWP/BMWP (SP) values exceeding
compared to the official   national  field protocols for the
100 were assigned high ecological status.
IBGN index commonly applied in France (Table 1). Data were
For the SPEAR index, new class boundaries were derived
checked for departure from the normal distribution using the
from the study of Liess and Von der Ohe3 to allow for the
Kolmogoroff Smirnoff test prior to analysis. The means of the
classification of ecological status of the selected reference sites.
metric values for each reference site were compared between
In the study of Liess and Von der Ohe,3 a sensitivity threshold
basins using one-way analysis of variance (ANOVA). Dun-
0.36 (median of species sensitivity) was set to classify half of
nett s multiple comparison test was employed to detect sig-
the determined taxa in that study to be at risk and the other half
nificant differences among means. For the Weser River basin,
to be not at risk. As the classification of species at risk did not
this comparison was also performed for the different stream
depend on any specific traits of species, a SPEAR metric value
types. To consider the relative variation of the mean index
of approximately 50% (according to half of the community)
values, the coefficient of variance (cv) was calculated.
was expected for communities at undisturbed reference sites.
The   predictivityHES  referred to the ratio of the number of
Interestingly, a mean metric value of 50% ( 7% s.d.) was
sites whose computed mean index values fell into the class
observed for the four non-contaminated sites (TU(D. magna) boundaries of the high ecological status compared to the total
values were o 4) of the previous field study.3 Therefore, a
number of reference sites. It was computed to quantify the
mean SPEAR value of 50% was set as centre of the high
index s power to detect non-contaminated streams across
ecological status class, with equidistant class boundaries of
Europe. Accordingly, the proportion of sites whose computed
7% that resulted in a lower class boundary of Z 43. The
mean index values exceeded at least the class boundaries of the
class size of 14% should cover the natural variability of the
good ecological status were referred to as   predictivityGES  .
SPEAR metric values at reference conditions. This procedure
Intercorrelation among single metrics were determined with
allowed for the establishment of an upper anchor for setting
Pearson s product moment correlation coefficient r (significant
class boundaries and subsequently the identification of depar- correlations for r Z 0.60; a = 0.05) without standardization
tures from expected reference conditions that may be caused by
of the metric values. All statistical analyses were carried out
Table 1 Overview of the taxonomic level of determination of the national applied assessment method (AM), the number of reference sites (n) in
each of five European river basins (RB), as well as mean standard deviation of metric values and descriptive statistics of the assessment methods
Assessment method (AM)
River basin Country of National applied Taxonomic level
(RB) RB AM of AM n BMWPf BBI SI (DE) SPEAR
Llobregat Spain BMWP (SP) Family 5 99 12   53 8
Scorff France IBGNa Familyb 6 168 6 9.3 0.4 1.72 0.04 46 3
Scheldt Belgium BBI Genus 8 83 22 7.4 0.6  51 5
Weser Germany SI (DE) Species 11 66 21 9.3 0.4 1.95 0.21 52 5
Porvoonjoki Finland  Genusb 5 102 23 6.9 1.4 2.08 0.07 47 5
Descriptive statistics
Mean cv 19.2% 11.1% 5.3% 9.9%
Applicabilityc 5/5 3(4)/5 2(3)/5 5/5
PredictivityHES d 44% 18% 6% 94%
PredictivityGES e 76% 59% 100% 100%
a b
Calculation of IBGN was not available from the AQEM software and, therefore, not performed. Data available from Schafer et al.20 was
¨
c
mainly at genus level. Number of countries, where the metric could be applied to monitoring data (+field study of Schafer et al.).20 d Mean ratio
¨
e f
of correctly classified sites of high ecological status. Mean ratio of correctly classified sites of at least good ecological status. BMWP adapted for
the Iberian Peninsula and respective class boundaries used for Llobregat data set.
c
This journal is The Royal Society of Chemistry 2007 J. Environ. Monit., 2007, 9, 970 978 | 973
using SPSS version 7.5.1.35 Graphs were created with the However, no significant differences were found between the
SigmaPlot graphics package.36 reference sites of the three stream types in the Weser River
basin. For the BBI, significant differences were only observed
between the Scorff River basin and all other basins. For the
Results
SPEAR index, no significant differences between any of the
river basins or stream types were observed, indicating a similar
General applicability of the indices
classification for all reference conditions.
The taxonomic level required for calculating the SI (DE) was For the SI (DE), single river basin cv values ranged from 2%
the species level, and therefore it was only applicable to the at the Scorff River basin and 3% in the Porvoonjoki River
Weser and Porvoonjoki River basins (Table 1). Although basin to 11% in the Weser River basin. The latter, somewhat
appropriate data for the Scorff River basin were available larger cv was mainly attributed to differences in stream
from own investigations,20 the taxonomic level of the official typology (see above). As can be seen from Table 1, the overall
monitoring programmes were limited to the family level. The cv that considered data from all basins was computed as 9%.
same applies to BBI that required information on the genus For the BMWP, the computed cv values ranged from 4% at
level, which was additionally available for the monitoring data the Scorff River basin to 32% at the Weser River basin, the
of the Scheldt River basin. The BMWP required family level latter, however, irrespective of the stream typology. This was
resolution and was therefore applicable to all datasets. The reflected in a high overall cv of 41% that stemmed from the
SPEAR index was also generally applicable as the calculation very high BMWP values in the Scorff River basin. For the
of the metric could be adapted to the required level of BBI, the cv values ranged from 5% at the Scorff River basin
taxonomic resolution of the national river monitoring and 8% at the Scheldt River basin to 20% at the Weser River
programmes. basin, again irrespective of the stream typology. The latter was
Besides the general applicability, the metric values of an
reflected in the overall cv of 16%. Finally, the cv values for the
index should be preferably similar for all river basins to allow
SPEAR index ranged from 6% in the Scorff River basin to
equal classification of water bodies on a European level. For 15% in the Llobregat River basin. The relatively low overall cv
the SI (DE), significant differences were observed between the of 11% was therefore comparable to the one of the SI (DE).
Scorff River basin and the Porvoonjoki River basin (Table 1).
Moreover, the metric values for river types no. 7 (mean SI =
Detection of non-contaminated sites
1.83 0.06), no. 14 (mean SI = 2.10 0.12) and no. 16 (mean
SI = 1.96 0.07) differed all significantly within the Weser For the SI (DE), the predictivityHES ranged from 0% for the
River basin (Fig. 1c). For the metric values of the BMWP, Scorff and for the Porvoonjoki River basins to only 10% for
significant differences between basins were found for the Scorff the Weser River basin. The correct classification of all remain-
River basin and all other basins. Moreover, the Weser River ing reference sites corresponded to high predictivityGES of
basin differed significantly from the Llobregat River basin as 90% to 100% (Fig. 1c). The predictivityHES for the BMWP
well as from the Porvoonjoki River basin (Table 1, Fig. 1a). ranged from 9% in the Weser River basin and 25% in the
Fig. 1 Box and whisker plots of mean metric values of all reference sites belonging to one of the five European river basins, separately for each of
the four examined indices: (a) scoring system of biological monitoring working group (BMWP), (b) Belgium biotic index (BBI), (c) German
saprobic index (SI (DE)) and (d) species at risk index (SPEAR). Class boundaries of good and high ecological status are included for comparison.
c
974 | J. Environ. Monit., 2007, 9, 970 978 This journal is The Royal Society of Chemistry 2007
Scheldt River basin to 100% in France. The remaining among the commonly used assessment methods of organic
reference sites were correctly classified as good ecological pollution ranged from r = 0.637 for the BBI vs. SI, r = 0.633
status, except for the Weser River basin. There, a low pre- for the BMWP vs. SI, to r = 0.742 for the BBI vs. BMWP, no
dictivityGES of only 27% yield 64% of the reference sites that significant correlations were found for the recently developed
missed the required quality objective of the WFD (Fig. 1a). SPEAR index to detect organic toxicants.
For the BBI, a predictivityHES of 100% was only achieved for
the Scorff River basin, whereas none of the reference sites in
Discussion
the other basins were classified accordingly. The predictivi-
tyGES ranged from only 40% for the Porvoonjoki River basin For the past decades, environmental policies in Europe ad-
to 75% for the Scheldt River basin. Hence, the remaining sites dressed the enhancement of ecosystem functions by focusing
were classified as possessing insufficient ecological status (Fig. on larger geographical areas. Related to freshwater systems,
1b). Finally, the SPEAR index classified all reference sites to the EU Water Framework Directive demanded for the protec-
have at least good ecological status. Single predictivityHES tion and management of European river basins in its totality
ranged from about 80% in the Porvoonjoki River basin as well by the EC Member States.1 Thus, indicators that are applic-
as for the Scorff River basin, to 100% in the other three able across national borders would be more desirable com-
river basins, resulting in an overall predictivityHES of 94% pared to several different assessment methods. In a study of
(Table 1). Gerhard et al.,37 the BMWP (SP) and the BBI were success-
fully applied to a Portuguese river, whereas Sandin and
Hering13 did the same with the saprobic index in rivers of
Correlation of assessment methods
Austria and the Czech Republic. However, in many countries,
The added value of an index of toxicity would be limited if it a single stressor (e.g. organic pollution) was regarded to be still
was highly correlated with the locally applied index of organic overwhelming,9 hence only a single metric was applied. In our
pollution. Therefore, pairwise scatterplots for all indices were study, this applied to the BBI and the BMWP (SP) that were
created (Fig. 2). The analysis of correlation revealed differ- employed in Belgium and Spain, respectively. According to the
ences between the three commonly applied assessment meth- Directive, though, it was not the impact of a single pressure on
ods and the SPEAR index. While the correlation coefficients individual BQEs that will determine ecological status
Fig. 2 Correlation of non-standardized metric values of three locally applied indices of organic pollution (e.g. increased BOD and nutrient
contents) and an adaptation of the recently developed SPEAR index. Linear regression lines are plotted for visualization of trends.
c
This journal is The Royal Society of Chemistry 2007 J. Environ. Monit., 2007, 9, 970 978 | 975
classification but the deviation of the whole community from ecological status was confirmed by most indices and for most
undisturbed reference conditions.9 Deteriorations from this countries. However, for the BMWP, some reference sites in the
status might be caused by both various kinds of natural Weser River basin and for the BBI, some sites in the Scheldt,
disturbances and especially anthropogenic ones, such as or- Porvoonjoki and also in the Weser River basins were mis-
ganic pollution (including eutrophication), toxic pollution classified. For the Weser River basin, it was known that auto-
(including acidification) and habitat degradation.2 The rele- saprobic effects might occur due to high biomass of plant
vance of different stressors may also vary for different surface material in slow moving streams with fine particulate sedi-
water bodies within a river basin. In this context, projects like ments.41 This might lead to misclassification of eutrophication
MODELKEY aim to link observed toxicity gradients to metrics like the BMWP and BBI. However, the BMWP
observed responses in natural communities of different trophic classified 44% of the reference sites as high ecological status,
levels, using novel indicators for an integrated assessment. whereas merely 6% and 18% were classified accordingly by
In the seventies, the application of the saprobic index the SI (DE) and the BBI, respectively. SPEAR was the only
revealed deterioration of the water quality of many German index that classified all reference communities across Europe
streams resulting from high loads of organic pollution.38 above the threshold of good ecological status of which 94%
Appropriate measures, such as the installation of new and were classified as high ecological status. Reference sites, where
the technical improvement of existing wastewater treatment only the SPEAR metric indicated a good ecological status,
plants, apparently lessened the impact of organic pollution in might have been disturbed by increased organic loads, re-
recent decades.39,40 Consequently, new stressors, which have flected in the metric values of the indices of organic pollution.
been masked before, became more apparent.29 Since the Nevertheless, according to the SPEAR index, pesticides and
module   saprobity  from the German Assessment System other organic toxicants did not affect the investigated reference
Macrozoobenthos no longer gives cause for concern and the sites.
module   acidification  of rivers is only applicable for few
stream types, the module   general degradation  is often
General applicability
responsible for assessing water bodies deviation from good
ecological status.12 However, while this module summarises Of all metrics examined, the SI (DE) performed best with
the effects of different stressors, such as degradation of the regard to low cv values for different basins and a low overall
stream morphology, land use, pesticides and hormone equiva- cv, especially when considering stream typology. The SI (DE)
lent compounds, it does not allow to differentiate clearly metric values indicated significant differences between the
between stress resulting from structural degradation and the three stream types (see above), though the trend of the mean
effects of the other mentioned stressors. Hence, an index that SI (DE) metric value for the three different stream types
could distinguish between the effects of morphological degra- followed a similar trend of the respective class boundaries of
dation and the toxic effects of organic pollutants would be high ecological status. This result verified the differences in the
useful. predefined class boundaries.34 Interestingly, no significant
differences between stream types were observed using the other
Selection of reference sites metrics. However, with the SI (DE), only two of the five river
basins could be assessed based on the sampling data of the
From a total of 578 monitoring sites in five European river
national monitoring programmes, due to the required taxo-
basins, 34 reference sites that met the above-mentioned re-
nomic resolution that is often not available for European
quirements were selected, corresponding to 126 macroinverte-
monitoring data. Organic pollution metrics like the BMWP
brate samplings. These reference sites were assumed to be high
were most variable (overall cv of 19%) compared to the
quality river sites as they were without any agricultural
saprobic index, which was also found by Springe et al.42 This,
activities in their upstream catchments and with floodplains
again, demanded the need for intercalibration studies prior to
preferably covered with unmanaged forest, where the coarse
a general application. The BBI in turn showed a medium
woody debris has not been removed.2 Selection was based on
variation in the river basin cv values, except for the Weser
existing information available from topographical maps, as
River basin. This was, however, not attributed to a single
well as knowledge from local experts, which led to the
metric value, i.e. metric values did not differ significantly
classification of the best existing monitoring sites. In this
among the three stream types. This corresponded to an overall
context, the STAR project (EVK1-CT-2001-00089) revealed
cv (11%) in the medium range between the SI (DE) and the
that in many cases, e.g. some lowland stream types, no sites
BMWP. In fact, the BBI s mean metric values were quite
were available meeting all of their criteria for a reference site.19
similar, except for the Scorff River basin. However, its applic-
In our study, this applied to the Weser River basin, where only
ability was limited to three river basins due to the required
reference sites of small size stream types were available.
level of the taxonomic resolution (genus).
In contrast, the trait based SPEAR index could be adjusted
Good and high ecological status classification
to the taxonomic resolution of all data. A relatively high mean
For sites representing reference conditions, high metric values cv for the Llobregat River basin was due to a single reference
and respectively high ecological status were expected. How- site with a very high SPEAR metric value of 66%, otherwise
ever, as the WFD aims at achieving good ecological status of the single cv values were most comparable to the overall cv of
all water bodies, correct classification of reference conditions 10%. In this context, Statzner et al.43 showed that with respect
for the examined indices was based on this status. A good to the natural variability, traits were quite stable for least-
c
976 | J. Environ. Monit., 2007, 9, 970 978 This journal is The Royal Society of Chemistry 2007
impacted streams across Europe. This could be confirmed by values differed highly among basins. This, in turn, would
the fact that the SPEAR metric values of the selected reference demand the need for intensive intercalibration studies.
conditions did not differ significantly among the five In contrast, the SPEAR metric was the only indicator that
river basins and the three stream types in the Weser River was not correlated to the three indices of organic pollution.
basin. Moreover, it was generally applicable to all monitoring data
Interestingly, the overall mean for all SPEAR metric values and at the same time correctly predicted high ecological status
of the 34 reference sites equalled 50%, which implicitly of non-contaminated streams. The latter is a crucial prerequi-
proved the selected threshold of 0.36 (ref. 3) also for the site to indicate departures from expected reference conditions.
determined taxa in this study. Compared to other indices, These results support the potential of the SPEAR index to be
SPEAR was not derived from empirical observations, like used as an indicator of organic toxicity. In two previous
e.g. the German saprobic index,29 but from trait based studies, the SPEAR index showed high correlation to the
considerations of the physiological sensitivity of aquatic measured gradients of toxicity,3,20 with metric values far below
invertebrates31 and additional life-history information.3 It the ones observed at the selected reference sites. In the context
is a relative measure of the fraction of the sensitive species of the WFD, the SPEAR index met the requirements of an
abundance and hence, more or less independent from the index concerning both the composition and abundance of the
taxonomic resolution of the applied sampling method. More- biotic assemblages.
over, in contrast to other species-based indices like the SI
(DE), all species of a community are considered, which
Conclusion
makes the SPEAR index more robust.
Based upon previous results and this study, the SPEAR
concept seems to be a promising assessment method to
Potential indices of toxicity
indicate organic toxicity. Using the relative fraction of species
to be at risk of a certain stressor, this approach may also be
Assessment methods that correctly indicate degradation due to
extrapolated to other BQEs and stresses. The authors propose
certain stressors are required to provide the necessary and
to use the locally developed eutrophication indices and the
relevant information to support decision-making on restoration
SPEAR metric simultaneously to extend the indication of
efforts to be spent or saved. Many of the existing national river
potential stressors that cause deviations from the good ecolo-
monitoring programmes, however, were originally designed to
gical status. This would allow for an integrated assessment on
monitor the potential impacts arising from organic pollution and
a river basin scale in compliance with the CIS of the WFD.
to classify streams according to their saprobic level. In a revision
However, further investigations on the interactions between
of the German saprobic index, Friedrich and Herbst29 stated the
stressors and an intercalibration study might be necessary
need for new specific indices in the context of the requirements of
prior to usage at the European level.
the WFD that go beyond the scope of the saprobic systems. At
the same time, for most BQEs there are no indicators addressing
the effects of organic toxicants. In this context, Liess and Ohe3 as
Acknowledgements
well as Schafer et al.20 stated a paucity of field investigations on
¨
The presented work was supported by the European Commis-
organic toxicants encompassing more than one river or stream.
sion through the Integrated Project MODELKEY (Contract-
Due to differences in the level of taxonomic resolution of the
No. 511237-GOCE). The field work studies, performed by
official monitoring programmes, the use of appropriate indices
Ralf B. Scha¨ fer were supported by the European Commission
might be limited if they could not be generally applied. For
through the HAIR Project (Contract-No. SSPE-CT-2003-
example, the SI (DE) seemed to be an appropriate indicator that
501997) and through a scholarship of the Studienstiftung des
assigned good ecological status to all selected reference sites.
deutschen Volkes e.V. (Bonn, Germany). We would like to
However, the index s requirement in terms of taxonomic data
thank Leila I. Cruzat and two anonymous reviewers for their
resolution compromised its applicability in Europe. Moreover,
useful comments on an earlier version of the manuscript.
significant differences between stream types would demand for
intercalibration studies, as respective class boundaries have to be
References
derived.11 In addition, relatively high correlation with other
indices of organic pollution were stated recently,9 which were
1 European Commission, Directive 2000/60/EC of the European
confirmed in this study. Parliament and of the Council of 23 October 2000 establishing a
framework for Community action in the field of water policy, Off.
The same held for the BBI and the BMWP that were highly
J. Eur. Comminities, L327, 22.12.2000, p. 77.
correlated (r = 0.742, p r 0.001). This was surprising, because
2 M. Furse, D. Hering, O. Moog, P. Verdonschot, R. K. Johnson, K.
the investigated sites were selected to reflect reference
Brabec, K. Gritzalis, A. Buffagni, P. Pinto, N. Friberg, J. Murray-
conditions and, therefore, the expected range of the metric Bligh, J. Kokes, R. Alber, P. Usseglio-Polatera, P. Haase, R.
Sweeting, B. Bis, K. Szoszkiewicz, H. Soszka, G. Springe, F.
values was rather limited. While similar classification of sites
Sporka and I. Krno, Hydrobiologia, 2006, 566, 3 29.
was desirable for indices applied in different Member States,
3 M. Liess and P. C. Von der Ohe, Environ. Toxicol. Chem., 2005, 24,
this fact was limiting the value of the simultaneous application
954 965.
4 R. K. Johnson, Ecol. Appl., 1998, 8, 61 70.
to indicate different stressors. For the BBI, the classification of
5 L. Sandin, PhD thesis, Silvestria 172, Swedish University of
reference sites that missed the good ecological status was most
Agricultural Sciences, 2001.
hindering for its potential use to detect toxicity gradients. The
6 R. Kolkwitz and M. Marsson, Mitt. K. Prüfungsanst. Wasserver-
BMWP, again, was applicable to all river basins, but metric sorg. Abwasserbeseit. Berlin, 1902, 1, 33 72.
c
This journal is The Royal Society of Chemistry 2007 J. Environ. Monit., 2007, 9, 970 978 | 977
7 V. Sladecek, System of water quality from the biological point of Quality, Ecological Assessment and Control, ed. P. J. Newman, M.
view, Arch. Hydrobiol., Bieh. Ergebn. Limnol., 1973, 7, 1 218. A. Piavaux and R. A. Sweeting, Commission of the European
8 J. L. Metcalfe-Schmith, in Biological water quality assessment of Communities, EUR 14606 En-Fr, 217 248, 1992.
rivers: use of macroinverterbrate communities, ed. P. Calow and G. 27 P. Rolauffs, D. Hering, M. Sommerhauser, S. Rodiger and S.
¨ ¨
E. Petts, The rivers handbook, Blackwell scientific publications, Jahnig, Entwicklung eines leitbildorientierten Saprobienindexes für
¨
London, 1994, vol. II, pp. 144 177. die biologische Fließgewässerbewertung, UBA Texte, 11/03, Bunde-
9 S. Birk and D. Hering, Hydrobiologia, 2006, 566, 401 415. sumweltamt, Berlin, 2003, p. 137.
10 R. A. E. Knoben, C. Ross and M. C. van Oirschot, Biological 28 C. Meier, J. Bohmer, R. Biss, C. Feld, P. Haase, A. Lorenz, C.
¨
assessment methods of watercourses, UN/ECE Task Force on Rawer-Jost, P. Rolauffs, K. Schindehutte, F. Scholl, A. Sun-
¨ ¨
monitoring and Assessment, 1995, p. 86. dermann, A. Zenker and D. Hering, Weiterentwicklung und
11 CIS WG 2.A Ecological Status (ECOSTAT), Guidance on the Anpassung des nationalen Bewertungssystems für Makrozoo-
intercalibration process, agreed version of WG 2.A Ecological benthos an neue internationale. Vorgaben Abschlussbericht im
Status meeting held 7 8 October 2004 in Ispra, Version 4.1. 14, Auftrag des Umweltbundesamtes, 2006, unpublished manu-
October 2004, 2004, ECOSTAT, Ispra. script.
12 AQEM consortium, Manual for the application of the AQEM 29 G. Friedrich and V. Herbst, Acta Hydrochim. Hydrobiol., 2004, 32,
method, A comprehensive to assess European streams using 61 74.
benthic macroinvertebrates, developed for the purpose of the 30 D. Hering, O. Moog, L. Sandin and P. F. M. Verdonschot,
Water Framework Directive, Version 1.0, February 2002. Hydrobiologia, 2004, 516, 1 20.
13 L. Sandin and D. Hering, Hydrobiologia, 2004, 516, 55 68. 31 P. C. Von der Ohe and M. Liess, Environ. Toxicol. Chem., 2004, 23,
14 C. A. Schriever, M. Hansler-Ball, C. Holmes, S. Maund and M. 150 156.
Liess, Environ. Toxicol. Chem., 2007, 26, 346 357. 32 T. N. Sherratt, G. Roberts, P. Williams, M. Whitfield, J. Biggs, N.
15 W. Brack, J. Bakker, E. de Deckere, C. Deerenberg, J. van Gils, M. Shillabeer and S. J. Maund, Environ. Toxicol. Chem., 1999, 18,
Hein, P. Jurajda, B. Kooijman, M. Lamoree, S. Lek, M. J. Lopez 2512 2518.
de Alda, A. Marcomini, I. Munoz, S. Rattei, H. Segner, K. 33 European Commission, Water Framework Directive (WFD) Com-
Thomas, P. C. von der Ohe, B. Westrich, D. de Zwart and M. mon Implementation Strategy Working Group 2.3 Reference
Schmitt-Jansen, Environ. Sci. Pollut. Res., 2005, 12, 252 256. conditions for inland surface waters (REF C-OND), Guidance
16 R. M. Hughes, in Defining acceptable status by comparing with on establishing reference conditions and Ecological Status class
reference conditions, ed. W. S. Davis and T. P. Simon, Biological boundaries for inland surface waters, final version, 30 April 2003,
Assessment and Criteria, Tools for Water Resource Planning and p. 86.
Decision Making, Lewis Publishers, Boca Raton, FL, 1995, pp. 34 U. Schmedtje, M. Sommerhauser, U. Braukmann, E. Briem, P.
¨
31 47. Haase and D. Hering, Top-down-bottom up-Konzept einer bio-
17 T. B. Reynoldson, R. H. Norris, V. H. Resh, K. E. Day and D. M. zonotisch begrundeten Fließgewassertypologie Deutschlands, in
¨ ¨ ¨
Rosenberg, J. N. Am. Benthol. Soc., 1997, 16, 833 852. Deutsche Gesellschaft für Limnologie (DGL), Tagungsbericht 2000
18 J. L. Stoddard, D. P. Larsen, C. P. Hawkins, R. K. Johnson and R. (Magdeburg), Tutzing, 2001, pp. 147 155.
H. Norris, Ecol. Appl., 2006, 16, 1267 1276. 35 SPSS for Windows, release 7.5.1, SPSS Inc., Chicago, IL, 1996.
19 R. C. Nijboer, R. K. Johnson, P. F. M. Verdonschot, M. Som- 36 Sigma Plot for Windows, release 4.0, SPSS Inc., Chicago, IL, 1997.
merhauser and A. Buffagni, Hydrobiologia, 2004, 516, 91 105. 37 A. Gerhardt, L. Janssens de Bisthoven and A. M. V. M. Soares,
¨
20 R. B. Schafer, T. Caquet, K. Siimes, R. Mueller, L. Lagadic and Environ. Pollut., 2004, 130, 263 274.
¨
M. Liess, Effects of pesticides on community structure and eco- 38 Landerarbeitsgemeinschaft Wasser LAWA (ed), Die Gewassergu-
¨ ¨ ¨
system functions in agricultural streams of three biogeographical tekarte der Bundesrepublik Deutschland, Mainz, 1976.
regions in Europe, Sci. Total Environ., 2007, 382, 272 285. 39 Landerarbeitsgemeinschaft Wasser LAWA (ed), Gewasseratlas
¨ ¨
21 J. Illies, Limnofauna Europaea, A Compilation of the European der Bundesrepublik Deutschland Biologische Gewassergutekarte
¨ ¨
Freshwater Species with Emphasis on their Distribution and Ecology, 1995, Berlin, 1996.
G. Fischer, Jena, 2nd edn, 1978. 40 Landerarbeitsgemeinschaft Wasser LAWA (ed), Gewasseratlas
¨ ¨
22 C. Zamora-Munoz, C. E. Sainz-Cantero, A. Sanchez-Ortega and J. der Bundesrepublik Deutschland Biologische Gewassergutekarte
Ü ´ ´ ¨ ¨
Alba-Tercedor, Water Res., 1995, 29, 285 290. 2000, Kulturbuchverlag, Berlin, 2002.
23 J. Alba-Tercedor and A. Sanchez-Ortega, Limnetica, 1988, 4, 41 U. Braukmann, Zoozönologische und saprobiologische Beiträge zu
51 56. einer allgemeinen regionalen Bachtypologie, Schweizebart, Stutt-
24 P. D. Armitage, D. Moss, J. F. Wright and M. T. Furse, Water gart, 1997.
Res., 1983, 17, 333 347. 42 G. Springe, L. Sandin, A. Briede and A. Skuja, Hydrobiologia,
25 N. de Pauw and G. Vanhooren, Hydrobiologia, 1983, 100, 153 168. 2006, 566, 153 172.
26 N. de Pauw, P. F. Ghetti, D. P. Manzini and D. R. Spaggiari, 43 B. Statzner, P. Bady, S. Doledec and F. Scholl, Freshwater Biol.,
Biological assessment methods for running water, in River Water 2005, 50, 2136 2161.
c
978 | J. Environ. Monit., 2007, 9, 970 978 This journal is The Royal Society of Chemistry 2007