Assessing the ecological status in species-poor systems: A fish-based index for
Mediterranean Rivers (Guadiana River, SW Spain)
Virgilio Hermoso a,b,∗ , Miguel Clavero c,d , Francisco Blanco-Garrido e , José Prenda a,f
a
Departamento de Biología Ambiental y Salud Pública, Universidad de Huelva, Avda. Andalucía s/n, 21071 Huelva, Spain
The Ecology Centre, School of Biological Sciences, University of Queensland, St Lucia, Qld 4072, Australia
Grup d’Ecologia del Paisatge, Àrea de Biodiversitat, Centre Tecnològic Forestal de Catalunya, Pujada del Seminari s/n, 25280 Solsona, Spain
d
Departament de Ciències Ambientals, Universitat de Girona, Campus de Montilivi, 17071 Girona, Spain
e
Mediodes, Consultoría Ambiental y Paisajismo S.L. Bulevar Louis Pasteur, 1 Blq. 2-1-1, 29010 Malaga, Spain
f
CIECEM, Parque Dunar Matalascañas s/n. Almonte, Huelva, Spain
b
c
a b s t r a c t
Keywords: ANNA
model Bioassessment
Freshwater fish
Introduced species
Site-specific indices
Type-specific indices
The assessment of the ecological status of freshwater ecosystems is a key issue for many international
laws such as the Water Framework Directive (WFD) in light of the actual impoverished status of these
ecosystems. Different multimetric approaches have been successfully developed in different freshwater environments. However, multimetric indices are difficult to apply to Mediterranean rivers basins,
characterized by freshwater fish communities very low in species richness and support a high number
of endemics with generalist and opportunistic life strategies. Here we follow a site-specific approach
to develop an adaptation of the multimetric procedure usually used in Indices of Biotic Integrity (IBI),
resulting in the Index of Community Integrity (ICI). We modeled the presence–absence of 10 native
freshwater fish species from the Guadiana River basin using the Assessment of Nearest Neighbor Analyses (ANNA) model, to assess the deviation of the observed and expected community composition at
reference condition. ANNA, which is a multi-species model, allowed incorporating in the index some
rare species, though not all of them present in the basin. Deviations were transformed into probabilities
of belonging to a reference site and species by species measures were then integrated in a final score.
The use of presence–absence only data reduces potential errors associated with estimations of species’
abundance and its seasonal changes, which is especially important in Mediterranean environments. The
ICI was sensitive to both habitat degradation and the degree of dominance of non-native species within
the freshwater fish community (or biotic perturbation), which have been traditionally overlooked in IBIs,
but was unresponsive to natural sources of variation. Given the site-specific nature and the simplicity of
the index that we propose here, we expect it to be applicable to other Mediterranean basins for which
predictive models can be produced.
1. Introduction
Freshwater ecosystems are of great importance to human culture, welfare and development and the intense human utilization
has led them to a generalized poor conservation status (Malmqvist
and Rundle, 2002). As a consequence, extinction rates are larger
among freshwater organisms than among terrestrial ones (Ricciardi
and Rasmussen, 1999). At present many freshwater species have
become extinct or are highly endangered (Darwall et al., 2008), a
situation that is particularly worrying in rivers of arid and semi-arid
∗ Corresponding author at: The Ecology Centre, School of Biological Sciences, University of Queensland, Goddard Building, St Lucia, Qld 4072, Australia.
Tel.: +61 07 3365 2709.
E-mail address: virgilio.hermoso@gmail.com (V. Hermoso).
regions, such as Mediterranean bioclimatic areas (Collares-Pereira
and Cowx, 2004; Smith and Darwall, 2006). As a response to the
generalized degradation of freshwaters, the development of tools
to assess their ecological status has been a key issue in the management of freshwater ecosystems in the last decades. International
laws as the Clean Water Act in the United States or the European
Water Framework Directive (WFD; European Commission, 2000)
have been introduced in an attempt to address these problems,
requiring the protection and restoration of biological integrity as
part of water quality standards.
According to the WFD the status of all European rivers and
streams must be assessed and classified into five predefined levels of ecological status or quality classes (bad, poor, moderate,
good and high) based on four biotic elements (diatoms, macroinvertebrates, macrophytes and freshwater fish). The aim of this
bioassessment is to help diagnose potential problems causing the
Fig. 1. Location of the 241 sampled sites and water courses network of the Guadiana River basin.
deterioration of freshwater ecosystems and to guide objective management plans and corrective actions leading them to a good
ecological status by 2015.
Many efforts have been devoted to the development of efficient tools to measure the ecological status of freshwater systems
based on fish. Karr (1981) originally designed an Index of Biotic
Integrity (IBI) for rivers in the northeast United States, following a
multimetric approach, in which the final score, or biotic integrity,
was the result of multiple partial evaluations made through different parameters or metrics. This IBI was originally composed of
twelve metrics that reflected the important components of community ecology: taxonomic richness, habitat and trophic guild
composition, and individual health and abundance. Following this
multimetric approach many other indices have been developed and
adapted to specific environmental conditions and requirements
(see Roset et al., 2007 for a review on these indices). Two different trends can be distinguished among recent approaches: spatially
based or type-specific methods (Melcher et al., 2007; Schmutz et
al., 2007b) and site-specific methods (Oberdorff et al., 2002; Pont
et al., 2007). Both approaches are based on the Karr’s original multimetric concept and the reference condition approach (Hughes et
al., 1986; Reynoldson et al., 1997; Bailey et al., 1998), comparing
an ecosystem potentially impaired against a similar ecosystem in
the best possible condition (its maximum ecological potential or
reference conditions).
Type-specific and site-specific methods differ in the way they
find the reference conditions for a given site under evaluation.
Type-specific approaches rely on grouping techniques to cluster
reference sites in a set of homogeneous landscape or biological
groups. Then a given site only has to be compared against the
reference conditions of the group in which it is included, which
are normally referred as the biotic characteristics of the best preserved sites within each group. The site-specific approach does not
require any classification and it simply finds specific reference conditions for every new given site according to its environmental
characteristics. A site-specific index was the option selected for the
development of the European Fish Index (EFI) which arose from the
European FAME project (FAME, 2004b) and is currently applied to
most of north and central European rivers. On the other hand, typespecific methods such as RIVPACS (Clarke et al., 2003) or AUSRIVAS
(Simpson and Norris, 2000) have been developed elsewhere for
other taxonomic groups. These two methods directly compare the
observed and expected communities using presence–absence data,
instead of relying on metrics gathering different compositional or
ecological aspects, as multimetric approaches do. However these
methods have received little attention in freshwater fish indices
(Hawkins, 2006; Kennard et al., 2006a; Carlisle et al., 2008).
The EFI has not been validated for the Mediterranean region of
Europe (Pont et al., 2007) and little attention has been given to the
adaptation of and IBI for these rivers (Sostoa et al., 2003; Ferreira
et al., 2007; Magalhães et al., 2008; Benejam et al., 2008). Mediterranean fish communities in Europe share common problems with
streams under similar climatic conditions in other regions, which
make it difficult to develop traditional IBIs (Moyle and Randall,
1998). These differences include: a reduced number of species per
site, a high number of endemics per basin and high spatial and temporal changes in fish communities (Moyle and Randall, 1998; Moyle
and Marchetti, 1999; Kennard et al., 2006a). Moreover, Mediterranean freshwater fish species have evolved in harsh environments
(e.g. facing severe droughts and floods) and have generally developed short lifespans, generalist habitat use, opportunistic feeding
strategies, high fecundity and early sexual maturity (e.g., VilaGispert and Moreno-Amich, 2002; Clavero et al., 2004). All of these
characteristics may impose serious limitations to the development
of an effective multimetric index for Mediterranean rivers, as it
did in other similar environments with even richer freshwater fish
communities (Shields et al., 1995; Harris and Silveira, 1999).
For these reasons we set out to develop a new approach suited
for the assessment of ecological the status of the Guadiana River
basin, which features a Mediterranean climate, using freshwater
fish communities. The base of the index proposed here is the assessment of the distance between the current community composition
and the reference condition, using the presence–absence of ten
native species in the basin. We tested its sensitivity to the most significant perturbation and natural gradients along the basin defined
through a Principal Component Analysis (PCA). We also propose
five quality classes according to the WFD needs and validated the
ICI evaluations comparing the output of others two local indices:
IBMWP and QBR, sensitive to different sources of habitat perturbation. The biotic status of native freshwater fish communities
was carried out as the focal output of the index, that traditionally
received little attention in these kinds of indices to biotic disturbances (when non-native species substitute natives) since even
though they suppose a major recognized threat to the conservation of freshwater fish, this perturbation has traditionally received
little attention in these kinds of indices.
Table 1
Environmental and biotic variables used to characterize the sampled sites. *Denotes human-biotic potentially perturbed variables and used to describe stressor gradients.
Mean values for the reference and the test data set are given separately. The total range is also shown.
Scale
Variable
Method
Mean reference
Mean test
Site
Water depth (cm)
Refuge cover availability (m2 of shelter/river width)
Elevation (m)
Relative position (dist. to the most headwater point/total length of the stream)
Stream order (Strahler)
Distance to headwater (km)
Distance to Guadiana River (km)
River width (m)*
Substrate coarseness (Wentworth scale)*
Riparian quality index (QBR, Munné et al., 2003)*
In situ
In situ
GIS
GIS
GIS
GIS
GIS
In situ
In situ
In situ
In situ
In situ
In situ
In situ
In situ
In situ
In situ
In situ
In situ
GIS
GIS
GIS
GIS
GIS
In situ
In situ
40.2
5.0
393.0
0.4
2.1
97.9
71.7
12.6
6.0
79.8
0.36
0.04
2.74
0.50
98.1
35.5
19.6
471.8
7.7
675.7
2033.8
15.7
83.8
99.6
1.0
12.0
40.0
5.9
392.7
0.5
2.1
63.2
52.7
10.4
4.9
54.3
1.78
0.13
4.63
1.20
123.8
72.7
20.9
684.9
7.8
554.0
2034.5
15.8
58.7
90.0
50.0
45.0
NH4 + (mg/L)*
NO2 − (mg/L)*
NO3 − (mg/L)*
PO5 3− (mg/L)*
SO4 2− (mg/L)*
Cl− (mg/L)*
Water temperature (◦ C)*
Conductivity (ES/cm)*
pH*
Annual precipitation (mm/m2 )
Solar radiation (10 kJ/m2 *dia*Em)
Average annual air temperature (◦ C)
Distance to the nearest reservoir upstream (km)*
Distance to the nearest reservoir downstream (km)*
% Exotic fish abundance*
% Exotic fish species richness*
Reach (500 m)
Basin
Range
7.0–200
0.0–60.6
7.1–974.9
0.04–1.00
1.0–6.0
3.6–1036.1
0.0–196.0
1.4–123.0
1.0–9.0
0–100
0.02–51.60
0.01–2.00
0.50–55.90
0.05–23.20
10.0–2380.0
2.0–834.0
9.4–32.6
38.0–3230.0
2.21–10.63
370.2–1114.5
1646–2227
13.0–18.0
0.0–196.0
0.2–115.8
0–100
0–100
Slope (0 /00 )
Sinuosity
Land uses
Urban/industrial (%)*
Intensive agriculture (%)*
Extensive agriculture (%)*
Natural (%)*
GIS
GIS
8.9
1.2
4.6
1.2
0.00–58.03
1.00–2.79
GIS
GIS
GIS
GIS
5.0
9.0
3.0
86.0
10.0
36.0
8.0
54.3
0.0–36.0
0.0–100.0
0.0–100.0
0.0–100.0
Basin area (drainage surface upstrema of each site, 103 km2 )
Gravelius index
Land uses
Urban/industrial (%)*
Intensive agriculture (%)*
Extensive agriculture (%)*
Natural (%)*
Reservoir (%)*
Population density (inhabitant/km2 )*
GIS
GIS
4.1
1.6
2.3
1.7
0.9–5919.1
1.14–2.68
GIS
GIS
GIS
GIS
GIS
GIS
4.1
13.1
7.8
78.5
0.15
15.6
4.6
27.0
12.5
59.6
0.43
23.6
0.0–6.7
0.0–97.0
0.0–89.1
0.9–100.0
0.0–21.2
0.0–459.3
2. Methods
2.1. Study area
The Guadiana River basin is located in the South-Western
Iberian Peninsula draining a total area of 67,039 km2 to the Atlantic
Ocean (Fig. 1). It features a typical Mediterranean climate, with
high intra and interannual discharge variation, with severe floods
and droughts (Gasith and Resh, 1999). Mean air temperature
ranges from 13 to 18.1 ◦ C, with a strong intra-annual variation in
extreme temperatures. Mean annual precipitation ranges from 350
to 1200 mm (with a mean of 450 mm) (Ninyerola et al., 2005).
Although it is a sparsely populated area (20 inhabitant/km2 ),
the landscape has been heavily transformed during the last century by agricultural activities. Nearly one half of the basin (49.1%)
is currently being used for agriculture (30.6% occupied with intensive agriculture as irrigated agriculture and 18.5% is occupied with
other agriculture, like olive groves or fruit trees). Due to the large
agricultural water demand, about 13,000 hm3 of water is regulated
by 88 big reservoirs (>1 hm3 ) and more than 200 smaller reservoirs
(<1 hm3 ). Water abstraction in wells is also a great problem in some
areas of the basin, since there are thousands of legal exploitations
and an uncertain number of illegal ones, with severe consequences
for the main aquifers. Other common human perturbations are
related to river channel modifications due to river channelization
and degradation and even complete depletion of the riparian forest
and water quality deterioration (Urrea and Sabater, 2009).
2.2. Fish community and habitat characterization
Fish communities were characterized in 241 localities across
the basin (Fig. 1), using electrofishing during spring (2005–2006).
Sampling was carried out through a single-pass at each location
without block-nets along 100 m long stretches whenever possible,
covering all habitats available at this scale following the recommendations of the FAME Consortium (2004b) for European rivers.
Sály et al. (2009) and Bertrand et al. (2006) demonstrated this
method adequately characterized spatial and temporal trends in
fish abundance and richness. Whenever used, abundances were
standardized as captures per unit of effort by the total length surveyed and time devoted [CPUE = number of fish/length (m) × time
(h)] (Blanco-Garrido et al., 2005).
Habitat was characterised by 38 environmental variables, covering three different spatial scales: site, reach and basin. Two
approaches were used in this characterization: (i) in situ measures, which described micro- and mesohabitat characteristics
at each sampling location; and (ii) remotely collected data,
using geographic information systems (GIS) to map broader scale
environmental attributes (Table 1). With the exception of physicochemical water characteristics, in situ variables were recorded from
transects located every 20 m within the surveyed river stretch
(9–27 measures/reach) and mean values were used for the analysis. Climatic variables were extracted from the Digital Climatic
Map of the Iberian Peninsula (Ninyerola et al., 2005) which was
built on long temporal series (15–50 years long), assuming that
data represented a mean climatic year in the area. We used a digital map provided by the Guadiana basin’s management authority
(Confederación Hidrográfica del Guadiana) to represent existing
land-cover. Potential changes in land uses were ruled out in the
short period of time in which the study was carried out. All
these environmental variables fell within two categories: (a) natural habitat variability in the basin and (b) human perturbation
(Table 1). All variables were checked for normality and transformed
when necessary prior to analysis (arcsine for land-cover variables
– expressed as proportion – and log (x + 1) for all other variables).
The database was divided into two independent sub-sets: (1) a
reference data set used to build and calibrate a predictive model and
(2) a test data set used in conjunction with the reference to establish quality classes according to the WFD exigencies. Reference sites
were defined following the approach proposed by Hermoso et al.
(2009a), and were characterised by low urban or agricultural land
uses within the whole basin and at the reach scale (500 m around
the sampling point) (Table 1). Furthermore, bank and channel structure as well as the riparian zone should be in natural condition (see
Hermoso et al., 2009a for more details on the definition of reference
sites in this area). To ensure freshwater fish communities at reference sites were not seriously impacted by introduced fish, all sites
where non-native species accounted for more than 5% of total fish
abundance were also discarded (Kennard et al., 2006b). Since the
number of reference sites identified when following these criteria
(p = 55 sites) was not sufficient for both model construction and validation, some reference sites were also chosen following the same
approach from adjacent basins in the same biogeographical region
(Tinto, Odiel and Guadalquivir basins, n = 35). These basins share
most of their native species and fell within the same biogeographic
region according to Doadrio (1988) and Abell et al. (2008). To avoid
the effect of basin species distinctiveness in our predictions we also
introduced the variable basin as an additional predictor.
2.3. Development of the Index of Community Integrity
The Index of Community Integrity (ICI) measures the deviation of the observed community composition from an expected
community following the reference condition approach (Hughes
et al., 1986; Reynoldson et al., 1997; Bailey et al., 1998). To do so
our index uses species composition at reference localities to build
an n-dimensional ordination space. Then the biotic ordination is
related to environmental characteristics using discriminant function analyses, so each new evaluated site can be spatially allocated
in the n-dimensional ordination space according to its environmental variables. Then it finds the most environmentally similar
reference sites in the ordination space and compares the observed
community at the site under evaluation with those close reference neighbours. Finally, ICI summarizes the partial evaluations
obtained in a species by species fashion (Fig. 2).
We first used reference sites to build and validate a predictive
model describing the presence–absence of the 10 most frequent
native species (Table 2) in relation to a set of environmental variables not affected by human perturbations (Table 1). This set of
non-perturbed environmental variables were used to match test
sites with the most similar environmental reference sites allowing site-specific predictions of expected taxonomic composition.
The predictions of the expected probabilities of occurrence of each
species were derived through an Assessment by Nearest Neighbour
Fig. 2. Flowchart for the development of Community Health Index. An example
of the transformation of the standardized residuals into sites’ probabilities to be in
reference condition is showed. The accumulated probability for a low residual (−1.6)
is very low (5%) while for a large one (1.6) is high (95%). Dotted lines separate steps
where the whole community (all species) or partial evaluations (species by species)
are used.
Analyses (ANNA) predictive model (Linke et al., 2005) which was
available for the region (see Hermoso et al., 2009a). In ANNA models, sites are treated as a continuum and predictions are derived
from the most environmentally similar reference sites, avoiding
any previous classification. It creates an n-dimensional ordination
space based on species composition (presence–absence) at reference sites and uses discriminant functions to allocate new sites
within the ordination space using environmental variables. The
model then finds the set of reference sites most environmentally
similar to the target site under evaluation and predicts its community composition based on the community composition of those
nearest neighbours (Linke et al., 2005). The contribution of each
selected reference site to the final prediction is weighted according
to their relative environmental distance to the test site. Then the
closer the selected reference site, the higher is its weight on predictions. Our ANNA model used the nearest six nearest reference
sites to predict species occurrences. This approach is a whole community predictive model, such as RIVPACS (Wright, 1995), allowing
Table 2
List of the 14 native freshwater fish found in the Guadiana River basin surveys. Only species with a prevalence >5% were used for the construction of the ICI. The 10 introduced
species found in the Guadiana’s river network and their prevalence are also shown. Prevalence is expressed as the % of sites where each species occurred (n = 241 sites).
Species
Author
Distribution
Prevalence (%)
Iberocypris alburnoides
Cobitis paludica
Squalius pyrenaicus
Luciobarbus microcephaus
Iberochondrostoma lemmingii
Luciobarbus comizo
Pseudochondrostoma willkommii
Salaria fluviatilis
Luciobarbus sclateri
Anaecypris hispanica
Gobio Lozanoi
Luciobarbus guiraonis
Anguilla anguilla
Alosa alosa
Lepomis gibbosus
Gambusia holbrooki
Micropterus salmoides
Cyprinus carpio
Esox lucius
Ameiurus melas
Alburnus alburnus
Carassius auratus
Australoheros facetum
Rutilus rutilus
Steindachner (1866)
Buen (1930)
Günter (1868)
Almaça (1967)
Steindachner (1866)
Steindachner (1864)
Steindachner (1866)
Asso (1801)
Günter (1868)
Steindachner (1866)
Doadrio and Madeira (2004)
Steindachner (1866)
Linnaeus (1758)
Linnaeus (1758)
Linnaeus (1758)
Girard (1859)
Lacepède (1802)
Linnaeus (1758)
Linnaeus (1758)
Rafinesque (1820)
Linnaeus (1758)
Linnaeus (1758)
Jenyns (1842)
Linnaeus (1758)
Iberian Peninslua
Iberian Peninsula
Iberian Peninsula
Guadiana
Iberian Peninsula
Iberian Peninsula
Iberian Peninsula
Circunmediterranean
Iberian Peninsula
Guadiana
Iberian Peninsula
Iberian Peninsula
North Atlantic
Eastern Atlantic
Non-native
Non-native
Non-native
Non-native
Non-native
Non-native
Non-native
Non-native
Non-native
Non-native
45
44
22
21
18
16
12
9
7
5
2
1
<1
<1
36
28
12
5
4
4
2
2
<1
<1
the prediction of rare species which should be discarded in other
traditional predictive methods. The ANNA model was built on a set
of 70 randomly chosen reference sites and validated through different tests using the remaining 20 reference sites (see Supplementary
materials for more details on model validation).
The basic unit used to construct the index was the probability of
a given site to be in reference condition. The probability was dependent on the deviation of an observed community composition
from a reference site with similar environmental characteristics (expected composition). We measured the deviation of the
observed presences-absences against the expected probabilities
in the best ecological potential status (O-E). We did this for each
species in each evaluated site (referred to as residuals hereafter).
Then, since the index compares the community composition in a
species by species fashion, we had ten different residuals for any
given site. Negative values indicate species loss (the species was
predicted to be present with a certain probability but it was absent).
The lower the residuals, the higher the probability of presence
unconfirmed hence. In the opposite extreme, positive residuals owe
to observed presences with low predicted probabilities. All residuals were standardized to a (0, 1) normal distribution (x-mean/SD
in the reference data set) and then transformed into probabilities of belonging to a reference site, as a way of rendering them
comparable (Fig. 2). Since we assumed that residuals decreased
with disturbance, we only used one-tailed evaluations for probabilities’ calculations. With this approach we assumed all the 10 native
species to be sensitive to any source of perturbation according to
previous studies on species tolerances in the same area (Hermoso
et al., 2009b). In this study the authors showed all these species to
be sensitive to particular sources of perturbation in the Guadiana
River basin. The inclusion of this set of species enabled us to evaluate a wide range of perturbations and to make diagnoses through
post hoc analyses, as it usually can be done with traditional IBI’s
parameters (using species-specific sensitivities). The probability of
a site to be in reference condition having a standardized residual x is
obtained from the cumulative normal distribution function corresponding to that x value [pnorm(x), according to Pont et al. (2007)].
Each species’ measure was then summed up in the final index score.
It ranged between 0 (the site has a null probability to be a reference site according to all the partial species’ evaluations) and 10
(the site has a high probability to be a reference site according to
all the partial species’ evaluations) (Fig. 2).
2.4. Scoring and validating the Index
To validate the index and to establish the cut-off points
between different quality classes according to the WFD exigencies, we checked its relationships with impairment gradients and
its response vs. other accepted indices.
A Principal Component Analysis (PCA) was carried out on the
environmental matrix (Table 1) to identify a set of independent
gradients representative of the main sources of variation in the
study area. PCA is usually used to summarize large datasets in a
number of orthogonal Principal Components (PCs) which gather the
main sources of variation within the original data. The first three
Principal Components (PC) accounted for 37% of the original variance recording a longitudinal headwater-mouth gradient (PC1), a
habitat impairment gradient (PC2) and a biotic perturbation gradient (PC3) (Table 3). This analysis highlighted the independence of
natural, habitat perturbation and biotic perturbation gradients.
We tested the general response of the index to the gradients described above and the evaluations of habitat perturbations
obtained using two alternative indices: (1) the Iberian Biomonitoring Working Party (IBMWP) which evaluates water quality based
on benthic macroinvertebrate communities (Alba-Tercedor et al.,
2002); and (2) the Qualitat del Bosc de Ribera index (QBR) used to
assess the perturbation status of the riparian-zone habitats (Munné
et al., 2003). We expected the ICI to be sensitive to perturbation
gradients and to show similar evaluations to the other two indices,
however, to be insensitive to natural sources of variation. In this
way, we could discard the effects of natural patterns from our evaluations.
The limits between quality classes were established to maximize the differences along the perturbation gradients and to reduce
the risk of committing Type I (inferring impairment when it does
not exist) and Type II (not detecting impairment when it does
exist) errors. To check the sensitivity of the ICI’s quality classes we
performed ANOVA analyses on the scores of the other two biotic
indices, using the current quality classes as categorical predictors.
With this analysis we did not aimed to intercalibrate the ICI and the
Table 3
Factor loadings for each variable in the first three Principal Components (PC) extracted from the Principal Component Analysis carried out in the whole habitat-biotic
variables data set listed in Table 1 except QBR. Only variables with loadings >0.6 in any PC are shown. % of explained variance (in parenthesis) and eigenvalues are also shown.
Denomination denotes the name for each PC in the text.
Variable
PC1 Alt (13.0%) 5.4
PC2 Alt (13.0%) 5.4
PC3 Alt (11.0%) 4.5
Denomination
Eigenv.
Longitudinal gradient
Habitat impairment
Biotic perturbation
Basin area
Distance to headwayer
River order
River width
Gravelius Index
Water depth
Reach Int. Agr.
Basin Int. Agr.
0.9
0.9
0.9
0.8
0.6
0.6
SO4 −
NO3 −
Substrate coarseness
Basin natural land
Precipitation
Reach natural land
% exotic richness
% exotic abundance
2
other indices (to ensure that a given locality will fall in the same
quality class when evaluated through different indices). Although
intercalibration is highly recommended to ensure comparability in
the evaluations obtained with different biotic indices (Sandin and
Hering, 2004; Birk and Hering, 2006) it was beyond the scope of this
study. As neither of both indices accounted for invasion status, we
also performed an additional check on the sensitivity of our quality
classes to the degree of biotic invasion (measured as the proportion
of non-native species richness and abundance).
The inclusion of rare species (with a probability of occurrence
<0.5 according to the ANNA model) in the ICI might have pernicious
effects on the index performance (Kennard et al., 2006a; Van Sickle
et al., 2007). To address this potential downfall we re-assessed the
index including only common species (expected to be present with
a probability >0.5). As the number of species included in the assessment varied among localities we led the final index score to a 0–10
scale (we calculated the average probability values obtained from
the species included and multiplied by10). We then checked the
sensitivity of this new approach as we did for the ICI (see above).
Finally, we explored the temporal consistency of the ICI. A set
of 58 of the 241 localities were re-sampled in spring 2007. These
sites covered a wide range of the ICI scores, ensuring reasonable
comparisons of different values between both periods. Moreover, a
year is a period of time short enough to discard major changes in the
ecological status due to human causes (except those due to drastic
events such as waste water spills). So we expected the index scores
derived from 2007 samples to be highly related to those obtained
from 2005 to 2006 data.
3. Results
The ICI ranged between 1.7 and 8.3 and clearly discriminated
between reference and test sites (t-test, t = 5.5 p < 0.001). The scores
were normally distributed in both cases (Kolmogorov–Smirnov
test; KS = 0.08, p > 0.2, KS = 0.07, p > 0.2 for the reference and the
test datasets, respectively). The index was sensitive to habitat and
biotic perturbations but insensitive to natural sources of variation
(Table 4). We also found strong correlations with the scores of the
other two indices measuring water quality and the perturbation
status of the riparian area (Table 4). The cut-off points fulfilled the
goals we demanded. The limits for the bad-poor and high-good
quality classes (Table 5) reduced the probability of Type I and II
errors minimizing the probability of labelling a reference site as
in bad condition and perturbed sites as in high condition. Different
−0.9
−0.9
−0.9
−0.8
−0.6
0.6
0.9
0.9
0.9
1.0
Table 4
Pearson correlation between natural and perturbation gradients summed up in
Table 3 and the Index of Community Integrity. Directions of association and p values
are shown (n = 241).
ICIall
Longitudinal gradient
Habitat impairment
Biotic perturbation
QBR
IBMWP
species
p = 0.613
p = 0.025 (−)
p < 0.001 (−)
p < 0.001 (+)
p < 0.001 (+)
ICIcommon species
p
p
p
p
p
= 0.008 (−)
= 0.040 (−)
< 0.001 (−)
< 0.001 (+)
< 0.001 (+)
QBR from Munné et al. (2003); IBMWP from Alba-Tercedor et al. (2002).
percentiles were used in both extremes to maximize the discriminatory power between quality classes. Mean index scores in the
reference data set were used as limit between moderate-good quality classes, ensuring that only sites with a score higher than an
average reference site could be labelled as in good condition, while
in moderate condition otherwise. Similarly the limit between poormoderate was set at the mean value of the perturbed sites. In this
way only sites in a condition higher than an average perturbed
site could be labelled as in moderate condition while in poor condition otherwise (Table 5). The ICI’s quality classes showed a high
concordance with the other two tested indices (Fig. 3). A previous analysis proved no redundancy between both indices and the
pressure gradients used to calibrate the present index (Pearson’s
r < |0.5| for all the relationships between the gradients in Table 3
and the QBR and IBMWP), avoiding circularity effects in the validation at this stage. There were significant differences in the scores
of both alternative indices among present quality classes (ANOVA,
F = 13.5, p < 0.001 for IBMWP and F = 4.6, p < 0.001 for QBR). Similarly, we found strong statistical differences in the proportion of
exotic species abundance and their richness among quality classes
(ANOVA; F = 22.37, p < 0.001 and F = 29.88, p < 0.001, respectively)
(Fig. 3). Moreover, the ICI was rather constant through two consecutive year periods (Pearson correlation; r = 0.6, p < 0.001). Finally,
the exclusion of rare species from the index (p < 0.5) had significant consequences, since the new scores became sensitive to the
natural gradient and responded less clearly to habitat perturbation
(Table 4).
4. Discussion
The reference condition approach assumes that human and
biotic impairment affects the local community composition, and
thus biological integrity is defined in terms of compositional sim-
Table 5
Quality classes for the Community Health Index. The cut-off points were established to maximize the index response to habitat and biotic impairment and reduce the risk
of Type I and II errors.
Quality class
Cut-off point
Aim
High
Percentile 95 perturbed sites
Less than 5% of perturbed sites are
misclassified as in high condition
(Reduced Type II error)
Mean reference sites
Only sites with a score higher than the
mean scores within sites in reference
condition are classified as Good
Mean perturbed sites
Only sites with a score higher than an
average perturbed site are classified as
moderate
Percentile 1 reference sites
Less than 1% of reference sites are
misclassified as in bad condition
(Reduced Type I error)
Good
Interval
6.46–10
5.51–6.46
Moderate
4.36–5.51
Poor
3–4.36
Bad
ilarity between observed and expected optimal situations in the
best ecological potential status (or reference condition) (Hughes
et al., 1986; Reynoldson et al., 1997; Bailey et al., 1998). Multimetric indices quantify the biological integrity through several
community attributes (as changes in guilds-based metrics) rather
than species by species comparisons. Since taxa are the basic unit
of communities, alterations at the taxonomic composition level
may occur before changes at other structural levels (Norris and
Hawkins, 2000; Hawkins et al., 2000). Moreover, different members
of the same ecological aggregation (as different species included
within the same metric in multimetric indices) may not respond
in the same way to a given disturbance or show independent
responses to different types and degree of degradations (Thiollay,
1992; Lindermayer et al., 1999, 2000). This could make indices
insensitive to specific disturbances. Additional problems have been
linked to traditional multimetric approaches, such as circularity in
the selection of sensitive metrics to conform the final index or the
tendency to make Type I errors associated also to metrics selection
(Norris and Hawkins, 2000). It is also difficult to develop complex
indices with numerous metrics in rivers with a small number of
native species, such as the Mediterranean ones (Miller et al., 1988).
Finally, multimetric indices require a good knowledge on species
basic ecology aspects (Norris and Hawkins, 2000) but this information is often lacking for Mediterranean fish fauna and hence
most of the ecological classifications are based on expert judgment,
as has been highlighted in other areas (Kleynhans, 1999). Therefore, a predictive approach which considers species composition
instead of elaborated metrics seems a more efficient way to face
the assessment of the ecological status in this type of environment.
However, some predictive approaches based on species composition such as RIVPACS or AUSRIVAS (Simpson and Norris, 2000;
Clarke et al., 2003) simplify the complex effects of natural or perturbation induced changes on community composition through the
use the O/E ratios as a synthetic measure of community integrity.
Some important changes in ecosystem structure or function may
not be detected through this simplistic approach (Karr and Chu,
2000; Norris and Hawkins, 2000). To overcome this drawback, and
according to Pont et al. (2006), changes at the whole community
were considered in this study through species by species comparisons making the index more powerful and flexible.
We used presence–absence data as the basis for the comparison
between the observed and the expected communities. It could be
argued that this approach would be insensitive to many stressors,
because individual populations of some species can suffer a considerable degradation before going locally extinct. However, at the
assemblage level, presence–absence data appears to be sufficiently
robust to allow the detection of reasonably subtle differences
0–3
among sites (Hawkins et al., 2000) and has widely been used in
other common indices (Clarke et al., 2003). On the other hand, population densities are submitted to greater temporal (seasonal and
inter-annual) and spatial rates of change than presence–absence
data even under natural conditions in Mediterranean and other
similar harsh environments (Meffe and Minckley, 1987; Matthews
and Marsh-Matthews, 2003; Magalhães et al., 2007). In addition,
species occurrence has been showed to be more persistent than
abundance through natural dramatic climatic events such as severe
droughts or floods, which are frequent in Mediterranean environments (Magalhães et al., 2007). This temporal variability usually
forces researchers to validate their indices through time series
to account for the effect of these natural changes (Pont et al.,
2006; Kennard et al., 2006a; Collier, 2008). The characterization of
species’ abundances is also more complex and difficult to standardize than species occurrences, while little or no improvement of the
first over the second approach when analyzing species-habitat relationship has been highlighted (Lewin et al., 2009). Key issues related
to sampling methodologies, such as sampling effort and methods
used, have strong consequences on bioassessment results (Lenat,
1993; Reynolds et al., 2003). Moreover, this also has economic
implications for the implementation of bioassessment programs
(Joseph et al., 2006). Systems requiring more detailed data (such
as abundance vs presence–absence) are more expensive, while a
cost-efficient method is always desired (Schmutz et al., 2007a).
Thus, the use of presence–absence data supposes not only a considerable simplification for the implementation of the index but
also a way to overcome other weakness related to the use of abundance data. Nevertheless standardized sampling protocols should
be always used when characterizing fish populations to ensure a
correct application of presence–absence indices.
Summing partial evaluations from different metrics to produce
a final index score cannot be recommended unless it could be
demonstrated every single partial measure to varies in the same
direction and with the same magnitude of response to damage
(Norris and Hawkins, 2000). The present index fulfils this exigency since the residuals between the observed presences-absence
and the expected probabilities tend to decrease with humanbiotic impairment given that all the species included in the index
have been proved to be sensitive to different sources of perturbation (Hermoso et al., 2009b). Sensitive species will disappear at
perturbed sites deriving high negative residuals in case of being
predicted to be present while tolerant species will remain at
perturbed sites implying higher positive residuals. These partial
evaluations are then transformed into probabilities of a site being in
reference condition. Each species’ evaluations had the same weight
within the final index score since all their derived probabilities were
Fig. 3. Responses of the Index of Community Integrity quality classes detailed in
Table 5 to different indicators of water quality, physical habitat and biotic perturbations. IBMWP (Alba-Tercedor et al., 2002) measures water quality, QBR Munné
et al. (2003) evaluates the perturbation status of the riparian area. The responses
to the invasion status (measured as the proportion of invasive species richness and
abundance) are also shown.
summed up without any weighting to avoid the index being more
sensitive to particular perturbations. Even rare species with low
prevalence (5%) were successfully modelled using the ANNA model,
overcoming potential weaknesses of discarding important ecological information (Pont et al., 2007). A similar approach has been
developed and applied to rivers in South Africa (Kleynhans, 1999),
what enhances the possibilities of this approach to be adapted to
other geographical areas sharing common problems. Both aspects,
the use of all species data with no weighting and the wide range of
environmental conditions accounted for when using data from the
whole basin, ensures that the index responds to a broad range of
perturbations.
The ability to distinguish between natural and impairmentinduced changes in
community composition is critical in
bioassessment (Fausch et al., 1990; Hughes et al., 1986; Norris
and Hawkins, 2000). The lack of concordance of the present index
and the longitudinal upstream-downstream gradient was managed
through the predictive model which accounted for a substantial
portion of the spatial variability of species’ presence–absence making its scores independent of natural variations. The inclusion of
the whole pool of modelled species played an important role, since
when rare species were discarded form the index it lost its independence to natural changes. A site-specific approach was used in
this study following the river continuum concept (Vanote et al.,
1980) avoiding artificial classifications and their derived consequences on bioassessment. We found reasonably good responses
for all the sources of perturbation tested as well as for the other
two indices (IBMWP and QBR). This result not only validates the
present index but also supports and enhances its ability to detect
different sources of habitat impairment. In this way we have tested
the index to respond to water quality and physical habitat degradation, which are the main perturbations also evaluated in the
IBMWP and QBR indices (Alba-Tercedor et al., 2002; Munné et al.,
2003), and the main perturbations gathered in the habitat impairment gradient defined using PCA analysis. In this PCA analysis we
found the most significant sources of human impairment in the
basin to be related to agriculture development, resulting in water
quality (increase in nitrate and sulphates) and physical habitat
degradation (e.g., siltation). Given that the index computes differences in assemblage composition in a species by species fashion, it
will gather the independent responses of all the species included
in the index (lead by their specific sensitivities). The response
showed to the biotic disturbance measured as the relative dominance of exotic species within the whole fish community is also
remarkable. Exotics are considered one of the major threats to the
conservation of native fish communities (Kaufman, 1992; Godinho
and Ferreira, 2000; Clavero et al., 2004) and thus a main issue to
consider when assessing the ecological status (Pont et al., 2006).
However, the effect of introduced species on native fish communities have often not been considered in recent studies (Oberdorff et
al., 2002; Pont et al., 2007; Ferreira et al., 2007; Magalhães et al.,
2008). Although a relationship between habitat deterioration and
the degree of invasion of freshwater fish communities has been
noted (e.g. Ross et al., 2001), the presence and dominance of exotic
species may occur even in areas with good physico-chemical water
quality (Kennard et al., 2005; Light and Marchetti, 2007). Thus,
impacts of exotic species on the ecological integrity of communities
may not be shown by physico-chemical habitat quality measures,
making especially important for any index to be sensitive to biotic
disturbances. The present ICI is capable of detecting not only commonly measured habitat perturbations but also but perhaps more
important the biotic status of the native freshwater fish community.
Furthermore, the relative weight of different perturbation factors
can be assessed in a post hoc diagnostic using each species partial evaluations based on species-specific sensitivities (Hermoso et
al., 2009b) which should guide the development of efficient corrective plans to lead the assessed freshwater ecosystems to a good
ecological status as the WFD requires.
The adaptation of the present index to different areas (or even
to a broader extent) would not be an issue since the only requisite needed is the development of accurate predictive models for
local faunas. In this way, the ICI could be adapted and validated
in other Mediterranean basins that at the moment lack efficient
fish-based bioassessment tools. In addition, it would also be interesting to evaluate the performance of the index on other taxa, such
as macroinvertebrates. However, further research is necessary on
this field, to test whether this method is suitable for other Mediterranean basins and/or taxa. The site-specific approach overcomes
the limitations imposed by closed classifications of river types, in
which an index is only applicable to a river previously included
within the same type. However, given the high basin endemicity
of Mediterranean freshwater fish communities, a basin approach
seems to be the best option for site-specific predictive methods. A
multi-species predictive approach is highly recommended for other
Mediterranean basins due to the reduced distribution areas of some
endemics and the high constraints imposed by low species richness values, forcing the inclusion of rare species. The evaluation of
model’s performance plays an important role in avoiding potential
inaccuracies in the ecological status derived from prediction errors.
Acknowledgements
The fieldwork was financed by the Confederación Hidrográfica
del Guadiana through the project “Diseñ o y Explotación de la Red de
Control Biológico de la Cuenca del Río Guadiana” and the Ministry of
Science and Technology (REN2002-03513, CGL2005-02699). V. H.
held a pre-doctoral fellowship funded by the Spanish Ministry of
Education (MEC; AP-2004-1414) and M.C. from a Juan de la Cierva
contract also funded by MEC. We also thank URS Corporation for
their help with field works and providing the macroinvertebrate
data used in this paper.
References
Abell, R., Thieme, M.I., Revenga, C., Bryer, M., Kottelat, M., Bogutskaya, N., Coad, B.,
Mandrak, N., Contreras Balderas, S., Bussing, W., Stiassny, M.L.J., Skelton, P., Allen,
G.R., Unmack, P., Naseka, A., Ng, R., Sindorf, N., Robertson, J., Armijo, E., Higgins,
J.V., Heibel, T.J., Wikramanayake, E.-, Olson, D., López, H.I., Reis, R.E., Lundberg,
J.G., Sabaj Pérez, M.H., Petry, P., 2008. Freshwater ecoregions of the world: a new
map of biogeographic units for freshwater biodiversity conservation. Bioscience
58, 403–414.
Alba-Tercedor, J., Jáimez-Cuéllar, P., Álvarez, M., Avilés, J., Bonada, N., Casas, J., Mellado, A., Ortega, M., Pardo, I., Prat, N., Rieradevall, M., Robles, S., Sáinz-Cantero,
C.E., Sánchez-Ortega, A., Suárez, M.L., Toro, M., Vidal-Abarca, M.R., Vivas, S.,
Zamora-Muñoz, C., 2002. Caracterización del estado ecológico de ríos mediterráneos ibéricos mediante el índice IBMWP (antes BMWPı̌). Limnetica 21, 175–
185.
Bailey, R.C., Kennedy, M.G., Dervish, M.Z., Taylor, R.M., 1998. Biological assessment
of freshwater ecosystems using a reference approach: comparing predicted and
actual benthic invertebrate communities in Yukon streams. Freshwater Biology
39, 765–774.
Bertrand, K.N., Gido, K.B., Guy, C.S., 2006. An evaluation of single-pass versus
multiple-pass backpack electrofishing to estimate trends in species abundance
and richness in prairie streams. Transactions of the Kansas Academy of Science
109, 131–138.
Benejam, L., Aparicio, E., Vargas, M.J., Vila-Gispert, A., García-Berthou, E., 2008.
Assessing fish metrics and biotic índices in a Mediterranean stream: effects of
uncertain native status of fish. Hydrobiologia 603, 197–210.
Birk, S., Hering, D., 2006. Direct comparison of assessment methods using benthic macroinvertebrates: a contribution to the EU Water Framework Directive
intercalibration exercise. Hydrobiologia 56, 401–415.
Blanco-Garrido, F., Clavero, M., Prenda, J., 2005. Habitat partitioning within congeneric species in Mediterranean basins at different spatial scales. Journal of
Fish Biology 67, 279–280.
Carlisle, D.M., Hawkins, C.P., Meador, M.R., Potapova, M., Falcone, J., 2008. Biological assessment of Appalachian streams based on predictive models for fish,
macroinvertebrate, and diatom assemblages. Journal of the North American
Benthologica Society 27, 16–37.
Clarke, R.T, Wright, J.F., Furse, M.T., 2003. RIVPACS models for predicting the
expected macroinvertebrate fauna and assessing the ecological quality of rivers.
Ecological Modelling 160, 219–233.
Clavero, M., Blanco-Garrido, F., Prenda, J., 2004. Fish fauna in Iberian Mediterranean
basins: biodiversity, introduced species and damming impacts. Aquatic Conservation: Marine and Freshwater Ecosystems 14, 575–585.
Collares-Pereira, M.J., Cowx, I.G., 2004. The role of catchment scale environmental management in freshwater fish conservation. Fisheries Management and
Ecology 11, 303–312.
Collier, K.J., 2008. Temporal patterns in the stability, persistence and condition of
stream macroinvertebrate communities: relationship with catchment land-uses
and regional climate. Freshwater Biology 53, 603–616.
Darwall, W., Smith, K., Allen, D., Seddon, M., McGregor Reid, G., Clausnitzer, V.,
Kalkman, V., 2008. Freshwater biodiversity–a hidden resource under threat. In:
J.C. Vié, C. Hilton-Taylor, S.N. Stuart (Eds.), Wildlife in a Changing World—An
Analysis of the 2008 IUCN Red List of Threatened Species. IUCN, Gland,
pp. 43–53.
Doadrio, I., 1988. Delimitation of areas in the Iberian Peninsula on the basis of
freshwater fishes. Booner Zoolische Beiträge 39, 113–128.
European Commission, 2000. Directive 2000/60/EC of the European Council and of
the Council of 23 October 2000 establishing a framework for Community action
in the field of water policy. Official Journal of the European Communities, L327
1–72.
FAME Consortium, 2004b. Manual for Application of the European Fish Index-EFI.
Version 1.1, January 2005, p. 66.
Fausch, K.D., Lyons, J., Karr, J.R., Angermeier, P.L., 1990. Fish communities as indicator of environmental degradation. American Fisheries Society Symposium 8,
123–144.
Ferreira, M.T., Caiola, N., Casals, F., Oliveira, J.M., De Sostoa, A., 2007. Assessing perturbation of river fish communities in the Iberian Ecoregion. Fisheries Management
and Ecology 14, 519–530.
Gasith, A., Resh, V.H., 1999. Streams in Mediterranean climate regions: abiotic influences and biotic responses to predictable seasonal events. Annual Review of
Ecology and Systematics 30, 51–81.
Godinho, F.N., Ferreira, M.T., 2000. Composition of endemic fish assemblages in relation to exotic species and river regulation in a temperate stream. Biological
Invasions 2, 231–244.
Harris, J.H., Silveira, R., 1999. Large-scale assessments of river health using an Index
of Biotic Integrity with low-diversity fish communities. Freshwater Biology 41,
235–252.
Hawkins, C.P., 2006. Quantifying biological integrity by taxonomic completeness: its
utility in regional and global assessment. Ecological Applications 16, 1277–1294.
Hawkins, C.P., Norris, R.H., Hogue, J.N., Feminella, J.W., 2000. Development and use
of predictive models for assessing the biological integrity of streams. Ecological
Applications 10, 1456–1477.
Hermoso, V., Linke, S., Prenda, J., 2009a. Identifying priority sites for the conservation
of freshwater fish biodiversity in a Mediterranean basin with high degree of
threatened endemics. Hydrobiologia 623, 127–140.
Hermoso, V., Clavero, M., Blanco-Garrido, F., Prenda, J., 2009b. Assessing freshwater
fish sensitivity to different sources of perturbation in a Mediterranean basin.
Ecology of Freshwater Fish 18, 269–281.
Hughes, R.M., Larsen, D.P., Omernik, J.M., 1986. Regional reference sites: a method
for assessing stream potential. Environmental Management 10, 629–635.
Joseph, L.N., Field, S.A., Wilcox, C., Possingham, H.P., 2006. Presence–absence versus
abundance data for monitoring threatened species. Conservation Biology 20,
1679–1687.
Karr, J.R., 1981. Assessment of biotic integrity using fish communities. Fisheries 6,
21–27.
Karr, J.R., Chu, E.W., 2000. Sustaining living rivers. Hydrobiologia 422/423, 1–14.
Kaufman, L., 1992. Catastrophic changes in species rich freshwater ecosystems: The
lessons of Lake Victoria. BioScience 42, 846–858.
Kennard, M.J., Arthington, A.H., Pusey, B.J., Harch, B.D., 2005. Are alien fish a reliable
indicator of river health? Freshwater Biology 50, 174–193.
Kennard, M.J., Pusey, B.J., Arthington, A.H., Harch, B.D., Mackay, S.J., 2006a. Development and application of a predictive model of freshwater fish assemblage
composition to evaluate river health in eastern Australia. Hydrobiologia 572,
33–57.
Kennard, M.J., Harch, B.D., Pusey, B.J., Arthington, A.H., 2006b. Accurately defining the reference condition for summary biotic metrics: a comparison of four
approaches. Hydrobiologia 572, 151–170.
Kleynhans, C.J., 1999. The development of a fish index to assess the biological
integrity of South African rivers. Water S.A. 25, 265–278.
Lenat, D.R., 1993. A biotic index for the southeastern United States: derivation and
list of tolerance values, with criteria for assigning water-quality ratings. Journal
of the North American Benthological Society 12, 279–290.
Lewin, W.C, Freyhof, J., Huckstorf, V., Mehner, T., Wolter, C., 2009. When no catches
matter: Coping with zeros in environmental assessments. Ecological Indicators.,
doi:10.1016/j.ecolind.2009.09.006.
Light, T., Marchetti, M.P., 2007. Distinguishing between invasions and habitat
changes as drivers of diversity loss among Californiaı̌s freshwater fishes. Conservation Biology 21, 434–446.
Lindermayer, B.D., Cunningham, R.B., McCarthy, M.A., 1999. Landscape analysis of
the occurrence of arboreal marsupials in the montane ash forest of the Central
Highlands of Victoria, southern Australia. Biological Conservation 89, 83–92.
Lindermayer, D.B., Margules, C.R., Botkin, D.B., 2000. Indicators of biodiversity for
ecologically sustainable forest management. Conservation Biology 14, 941–950.
Linke, S., Norris, R.H., Faith, D.P., Stockwell, D., 2005. ANNA: A new prediction method
for bioassessment programs. Freshwater Biology 50, 147–158.
Magalhães, M.F., Beja, P., Schlosser, I.J., Collares-Pereira, M.J., 2007. Effects of multiyear droughts on fish assemblages of seasonally drying Mediterranean streams.
Freshwater Biology 52, 1494–1510.
Magalhães, M.F., Ramalho, C.E., Collares-Pereira, M.J., 2008. Assessing biotic integrity
in a Mediterranean watershed: development and evaluation of a fish-based
index. Fisheries Management and Ecology 15, 273–289.
Malmqvist, B., Rundle, S.R., 2002. Threats to the running water ecosystems of the
world. Environmental Conservation 29, 134–153.
Matthews, W.J., Marsh-Matthews, E., 2003. Effects of droughts on fish across axes
of space, time and ecological complexity. Freshwater Biology 48, 1232–1253.
Meffe, G.C., Minckley, W.L., 1987. Persistence and stability of fish and macroinvertebrate assemblages in a repeatedly disturbed Sonora desert stream. American
Midland Naturalist 117, 177–191.
Melcher, A., Schmutz, S., Haidvogl, G., Moder, K., 2007. Spatially based methods to
assess the ecological status of European fish assemblage types. Fisheries Management and Ecology 14, 453–463.
Miller, D.L., Leonard, P.M., Hughes, R.M., Karr, J.R., Moyle, P.B., Schrader, L.H., Thompson, B.A., Daniel, R.A., Fausch, K.D., Fitzhugh, G.A., Gammon, J.R., Halliwell, D.B.,
Angermeier, P.L., Orth, D.J., 1988. Regional applications of an index of biotic
integrity for use in water resource management. Fisheries 13, 12–20.
Moyle, P.B., Marchetti, M.P., 1999. Application of indices of biotic integrity to California streams and watersheds. In: T.P. Simon (Ed.), Assessing the Sustainability
and Biological Integrity of Water Resources Using Fish Communities. CRC Press,
Boca Raton, FL, pp. 367–380.
Moyle, P.B., Randall, P.J., 1998. Evaluating the biotic integrity of watersheds in the
Sierra Nevada, California. Conservation Biology 6, 1318–1326.
Munné, A., Prat, N., Solá, C., Bonada, N., Rieradevall, M., 2003. A simple field method
for assesing the ecological quality of riparian habitat in rivers and streams:
QBR index. Aquatic Conservation: Marine and Freshwater Ecosystems 13, 147–
163.
Ninyerola, M., Pons, X., Roure, J.M., 2005. Atlas Climático Digital de la Península
Ibérica. Metodología y aplicaciones en bioclimatología y geobotánica. Universidad Autónoma de Barcelona, Bellaterra, ISBN: 932860-8-7.
Norris, R.H., Hawkins, C.P., 2000. Monitoring river health. Hydrobiologia 435, 5–17.
Oberdorff, T., Pont, D., Hugueny, B., Porcher, J.P., 2002. Development and validation
of a fish-based index for the assessment of “river health” in France. Freshwater
Biology 47, 1720–1734.
Pont, D., Hugueny, B., Beier, U., Goffaux, D., Melcher, A., Noble, R., Rogers, C., Roset,
N., Schmutz, S., 2006. Assessing river biotic condition at a continental scale: a
European approach using functional metrics and fish assemblages. Journal of
Applied Ecology 43, 70–80.
Pont, D., Hugueny, B., Rogers, C., 2007. Development of a fish-based index for the
assessment of river health in Europe: the European Fish Index. Fisheries Management and Ecology 14, 427–439.
Reynolds, L., Herlihy, A.T., Kaufmann, P.R., Gregory, S.V., Hughes, R.M., 2003. Electrofishing effort requirements for assessing species richness and biotic integrity
n western Oregon streams. North American Journal of Fisheries Management
23, 450–461.
Reynoldson, T.B., Norris, R.H., Resh, V.H., Day, K.E., Rosenberg, D.M., 1997. The reference condition: a comparison of multimetric and multivariate approaches to
assess water-quality impairment using benthic macroinvertebrates. Journal of
the North American Benthological Society 16, 833–852.
Ricciardi, A., Rasmussen, J.B., 1999. Extinction rates of North American freshwater
fauna. Conservation Biology 13, 1220–1222.
Roset, N., Grenouillet, G., Goffaux, D., Pont, D., Kestemont, D., 2007. A review of
existing fish assemblage indicators and methodologies. Fisheries Management
and Ecology 14, 393–405.
Ross, R.M., Lellis, W., Bennet, R.M., Johnson, C., 2001. Landscape determinants of
nonindigenous fish invasions. Biological Invasions 3, 347–361.
Sály, P., Erós, T., Takács, P., Specziár, A., Kiss, I., Bíró, P., 2009. Assemblage level monitoring of stream fishes: The relative efficiency of single-pass vs. double-pass
electrofishing. Fisheries Research 99, 226–233.
Sandin, L., Hering, D., 2004. Comparing macroinvertebrate indices to detect organic
pollution across Europe: a contribution to the EC Water Framework Directive
intercalibration. Hydrobiologia 516, 55–68.
Schmutz, S., Cowx, I.G., Haidvolg, G., Pont, D., 2007a. Fish-based methods for assessing European running waters: a synthesis. Fisheries Management and Ecology
14, 369–380.
Schmutz, S., Melcher, A., Frangez, C., Haidvolg, G., Beier, U., Böhmer, J., Breine, J.,
Simoens, I., Caiola, N., De Sostoa, A., Ferreira, M.T., Oliveira, J., Grenouillet, G.,
Goffaux, De Leeuw, J.J., Noble, R.A.A., Roset, N., Virbickas, T., 2007b. Spatially
based methods to assess the ecological status of riverine fish assemblages in
European ecoregions. Fisheries Management and Ecology 14, 441–452.
Shields Jr., F.D., Knight, S.S., Cooper, C.M., 1995. Use of the index of biotic integrity to
assess the physical habitat degradation in warmwater streams. Hydrobiologia
312, 191–208.
Simpson, J.C., Norris, R.H., 2000. Biological assessment of river quality: development of AUSRIVAS models and outputs. In: Wright, J.F., Sutcliffe, D.W., Furse,
M.T. (Eds.), Assessing the biological quality of fresh waters; RIVPACS and other
techniques. Freshwater Biological Association, Ambleside, Cumbria, U.K, pp.
125–142.
Smith K.G., Darwall, W.R.T. (Eds.), 2006. The Status and Distribution of Freshwater
Fish Endemic to the Mediterranean Basin. IUCN, Gland, Switzerland/Cambridge,
UK, v+34 pp.
Sostoa, A., Casals, F., Caiola, N.M., Vinyoles, D., Sánchez, S., Franch, C., 2003.
Desenvolupament d’un índex d’integritat biòtica (IBICAT) basat en l’ús dels
peixos com a indicadors de la qualitat ambiental dels rius a Cataluñ a. Documents tècnics de l’Agència Catalana de l’Aigua, 203 pp. Available on-line at:
http://www.gencat.net/aca/en//planificacio/directiva/treballs.jsp#C .
Thiollay, J., 1992. The influence of selective logging on bird species diversity in a
Guianan rain forest. Conservation Biology 6, 47–63.
Urrea, G., Sabater, S., 2009. Epilitic diatom assemblages and their relationship to
environmental caharcteristics in an agricultural watershed (Guadiana River, SW
Spain). Ecological Indicators 9, 693–703.
Van Sickle, J., Larsen, D.P., Hawkins, C.P., 2007. Exclusion of rare taxa affects performance of the O/E index in bioassessment. Journal of the North American
Benthological Society 26, 319–331.
Vanote, R.L., Minshall, G.W., Cummins, K.W., Sedell, J.R., Cushing, C.E., 1980. The
river continuum concept. Canadian Journal of Fisheries and Aquatic Science 37,
130–137.
Vila-Gispert, A., Moreno-Amich, R., 2002. Life-history patterns of 25 species from
European freshwater fish communities. Environmental Biology of Fishes 65,
387–400.
Wright, J.F., 1995. Development and use of a system for predicting the macroinvertebrate fauna in flowing waters. Australian Journal of Ecology 20, 1881–2197.