Freshwater Biology (2006) 51, 1730–1743
doi:10.1111/j.1365-2427.2006.01602.x
APPLIED ISSUES
Fish community changes after minimum flow increase:
testing quantitative predictions in the Rhône River at
Pierre-Bénite, France
N I C O L A S L A M O U R O U X , * J E A N - M I C H E L O L I V I E R , † H E R V É C A P R A , * M A R C Z Y L B E R B L A T , ‡
ANDRE CHANDESRIS* AND PASCAL ROGER*
*CEMAGREF, Biologie des Ecosystèmes Aquatiques, Lyon, France
†
UMR CNRS 5023 Ecologie des Hydrosystèmes Fluviaux, Université Lyon 1, Villeurbanne, France
‡
Compagnie Nationale du Rhône, Lyon, France
SU M M A R Y
1. Many aspects of the flow regime influence the structure of stream communities, among
which the minimum discharge left in rivers has received particular attention. However,
instream habitat models predicting the ecological impacts of discharge management often
lack biological validation and spatial generality, particularly for large rivers with many
fish species.
2. The minimum flow at Pierre-Bénite, a reach of the Rhône river bypassed by artificial
channels, was increased from 10 to 100 m3 s)1 in August 2000 (natural mean discharge
1030 m3 s)1), resulting in a fivefold increase in average velocity at minimum flow. Fish
were electrofished in several habitat units on 12 surveys between 1995 and 2004.
3. Principal components analysis revealed a significant change in the relative abundance of
fish species. The relative abundance of species preferring fast-flowing and/or deep
microhabitats increased from two- to fourfold after minimum flow increase. A change in
community structure confirmed independent quantitative predictions of an instream
habitat model. This change was significantly linked to minimum flow increase, but not to
any other environmental variables describing high flows or temperature at key periods of
fish life cycle. The rapidity of the fish response compared with the lifespan of individual
species can be explained by a differential response of specific size classes.
4. The fish community at Pierre-Bénite is in a transitional stage and only continued
monitoring will indicate if the observed shift in community structure is perennial. We
expect that our case study will be compared with other predictive tests of the impacts of
flow restoration in large rivers, in the Rhône catchment and elsewhere.
Keywords: dams, flow regime, habitat restoration, instream habitat model, large rivers
Introduction
Many aspects of the flow regime such as minimum
flow, rate of flow change, flood frequencies and timing
Correspondence: Dr Nicolas Lamouroux, CEMAGREF,
UR Biologie des Ecosystèmes Aquatiques, 3 bis quai Chauveau,
CP 220, F-69336 Lyon Cedex 09, France.
E-mail: [email protected]
1730
influence the structure of freshwater communities
(Poff et al., 1997; Richter et al., 2003; Propst & Gido,
2004). Consequently, efforts to restore flow conditions
to benefit river ecosystem health are increasing worldwide (e.g. database of The Nature Conservancy,
http://www.nature.org/initiatives). Flow restoration
projects can be viewed as natural experiments that
should be designed, monitored and adapted through
time with a strong cooperation between scientists,
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd
Fish community and minimum flow
managers and water users (Poff et al., 2003). Cooperation among these actors is, however, complicated by
the high uncertainty of ecological models that predict
the impacts of restoration scenarios. In particular,
quantitative predictions of the ecological impacts of
changes in flow regime often lack biological validation
and spatial generality.
Minimum flow in rivers is one aspect of the flow
regime that has received particular attention in scientific studies and water regulations, because it has been
reduced by water abstraction and water diversion in
many rivers. Low flows can be considered as ‘bottleneck’ conditions that provide critical stresses or opportunities for a wide array of fish specific life stages (Poff
et al., 1997; Lamouroux et al., 1999c). Instream habitat
models became popular over the last three decades for
quantifying the impacts of minimum flow changes on
fish populations (Bovee, 1982; Gore & Nestler, 1988;
Lamouroux et al., 1999c). These models combine biological models that describe the preferences of fish in
terms of velocity, depth and particle size, and a
hydraulic model that estimates how the physical
habitat varies with discharge. Despite abundant use
of these models worldwide, the extent to which fish
communities actually respond to minimum flow restoration at a site is largely unpredictable. ‘Before–after’
tests of the impacts of minimum flow restoration have
generally focused on target populations (often salmonids) and yielded contrasting results (e.g. Harris,
Hubert & Wesche, 1991; Phillips, Ory & Talbot, 2000;
Jowett & Biggs, 2004; Souchon & Capra, 2004). We are
not aware of such ‘before–after’ tests for fish communities in large rivers. Fish communities in natural large
rivers are generally viewed as more diverse and less
constrained by abiotic factors than communities in
small streams. However, flow regulation in large rivers
strongly modifies the environmental conditions in
these systems, with potential consequences on community structure (Galat & Zweimüller, 2001).
In this paper, we analyse the fish community
variations at Pierre-Bénite, a reach of the Rhône river
bypassed by artificial channels (average natural
discharge 1030 m3 s)1), between 1995 and 2004. Minimum flow in the reach was increased from 10 to
100 m3 s)1 in 2000 as part of a national restoration
programme concerning the Rhône river. This case
study provides a preliminary analysis of the impacts
of minimum flow change on fish in a large regulated
river, where multiple species coexist. It is also a
1731
unique opportunity to test the predictability of the
changes according to an instream habitat model that
was calibrated before 2000 (Lamouroux et al., 1999c).
This model combined a hydraulic model of the reach
and models of fish hydraulic preference developed
elsewhere in the Rhône basin. The model, run
‘blindly’ at Pierre-Bénite before minimum flow
change, predicted higher relative densities of fish
species preferring fast and/or deep hydraulic conditions after minimum flow increase.
Methods
The Rhône river and the decennial restoration
programme
The Rhône river is the largest river flowing through
France (mean discharge >1700 m3 s)1 at the mouth). As
other large rivers in Europe, it has had a long history of
regulation and damming, especially during the 19th
(embankment for navigation and flooding issues) and
20th (dams built for hydropower and other water
needs) centuries (Bravard, 1987; Schiemer & Spindler,
1989; Fruget, 1992). Sixteen reaches of the French part
of the river, each of length between 2 and 30 km, are
bypassed by artificial channels. Minimum flows in
these bypassed reaches range between 1/300 and 1/5
of the natural mean flow. When the maximum discharge processed by the power plants is exceeded in
the artificial channels, bypassed reaches evacuate high
flows and undergo major changes in hydraulic conditions. Most bypassed reaches have retained a longitudinal morphology close to natural conditions with
sequences of pools, runs and riffles. Fish communities
are also more similar to pre-impoundment than communities of artificial channels and reservoirs (Persat,
1988). However, species adapted to the fast-flowing
and deep conditions of large flowing rivers (e.g.
grayling, barbel, dace or the non-native nase) have
suffered from the reduction in discharge (Persat, 1988;
Lamouroux et al., 1999c), as in other large regulated
systems in Europe and elsewhere (Galat & Zweimüller,
2001; Aarts, Van den Brink & Nienhuis, 2004).
Following a number of local and national initiatives,
a decennial restoration programme of the Rhône river
was implemented by the French government in 1998
with the general objective of restoring the ecological
attributes of a large running river. The programme
involves the rehabilitation of migratory pathways and
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
1732
N. Lamouroux et al.
connections with secondary channels (Amoros, 2001)
as well as the increase of minimum flow rates in some
bypassed reaches of the river. A close cooperation
between scientists, managers and water users was
organised as part of the decennial programme,
making it possible to define a common strategy for a
consistent monitoring and evaluation of the various
restoration operations (André & Olivier, 2003).
The bypassed reach at Pierre-Bénite
The bypassed reach at Pierre-Bénite is located 6 km
downstream of the city of Lyon (Fig. 1). The reach is
10 km long and includes several marked riffles.
Natural annual mean and minimum discharge rates
at Pierre-Bénite are 1030 and 300 m3 s)1, respectively.
At natural mean discharge the reach is 160 m wide
and 3.7 m deep. The Pierre-Bénite dam regulates flow
at the upstream end of the bypassed reach. The last
4 km of the bypassed reach are under the hydraulic
influence of another reservoir situated further downstream (Vaugris dam) and hence our study only deals
with the upstream, free-flowing 6 km of the reach
(Fig. 1). No important tributaries discharge into the
bypassed reach. The shorelines are partly embanked
and several dikes provide a diversity of habitat
conditions for juvenile fish along the banks (Nicolas
& Pont, 1997). Several parameters of water quality
(oxygen, conductivity, pH, nitrites, nitrates, ammonium, phosphates, carbonates and sulphates) were
measured up to five times a year during the study
(Lamouroux et al., 2004). All measurements were
rated as good according to national criteria and as
they did not change during our study they are not
considered here. Water temperature was recorded
daily in the bypassed section (except before 27 April
1995 and between 22 February 1999 and 06 May 1999)
and typically varies at Pierre-Bénite between 4 °C in
winter and 25 °C in summer (Fig. 2).
Before 2000, the minimum flow at Pierre-Bénite was
20 m3 s)1 between April and August and 10 m3 s)1
during the other months. In August 2000, minimum
flow was increased to 100 m3 s)1 and because of this
change average water depth at minimum flow doubled (from 1.2 to 2.1 m) and average velocity
increased fivefold (from 0.07 to 0.35 m s)1). Daily
discharge rates, recorded in the reach during the
study, confirm this change and reflect the typical flow
regime of the Rhône river below Lyon (Fig. 3). The
bypassed reach discharges high flows both in spring
(glacial influence from the Alpes) and autumn–winter
(pluvial influence), and discharges the minimum flow
between high flow events.
Fish sampling
Twelve fish surveys were made in the bypassed reach
between 1995 and 2004 at or around minimum flow;
six before and six after 2000. Eight of the surveys were
Daily temperature (°C)
30
25
20
15
10
5
01/01/05
01/01/04
01/01/03
01/01/02
01/01/01
01/01/00
01/01/99
01/01/98
01/01/97
01/01/96
01/01/95
0
Date
Fig. 1 The Rhône bypassed section at Pierre-Bénite.
Fig. 2 Daily water temperature at Pierre-Bénite.
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
Fish community and minimum flow
By-passed reach
1733
Total Rhône river
10 000
1996
1000
100
10
10 000
Daily discharge (m3 s–1, log-scale)
1999
1000
100
10
10 000
2001
1000
100
10
10 000
2003
1000
Fig. 3 Daily discharge in the bypassed
section (left column) and total daily discharge of the Rhône river (right column).
Shown for a dry (1996) and wet (1999)
year before minimum flow increase, and
for a wet (2001) and dry (2003) year after
minimum flow increase.
100
10
j
f m a m j
made in autumn (October 1995, September 1996,
September 1998, September 1999, October 2001, September 2002, September 2003 and October 2004), the
others in spring or early summer (July 1995, March
1996, June 2002 and June 2004). During each field
survey, fish were collected by electrofishing in 23–50
independent habitat units (total number ¼ 401; mean
surface area 36 m2; standard deviation 51 m2) using
an open-sampling technique adapted to large rivers
(Vadas & Orth, 1993; Lamouroux et al., 1999c; Thévenet & Statzner, 1999). On average, a fish survey at
Pierre-Bénite involved 33 habitat units of 36 m2, i.e. a
sampled surface area approximately equal to 0.2% of
the total surface area of the reach. Although the
efficiency of open electrofishing techniques in large
rivers is difficult to assess, Thévenet & Statzner (1999)
demonstrated that electrofishing habitat units of
j a s o n d
j
f m a m j
j a s o n d
Month
50 m2 was comparably efficient across species at
depths <1.5 m, independent of the habitat conditions.
Habitat units were sampled by wading when possible
or from a small boat (EFKO type FE48000, generator
300–400 V, engine 9.5 kW). Within each habitat unit,
fish were identified to species and measured. Habitat
units were randomly chosen in pools, runs, riffles, at
the channel centre or its margins and distributed
proportionally according to habitat availability. However, portions of the reach deeper than 2 m could not
be sampled because of the limits of electrofishing in
deep conditions. Hence, habitat units were not similar
between field surveys but were always selected with
the same stratified random strategy. On average
across surveys, habitat units were comparably distributed in pools, runs or riffles before and after
minimum flow increase (3%, 74% and 23%, respect-
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
1734
N. Lamouroux et al.
ively, before and 0%, 71% and 29%, respectively,
after 2000). Similarly, units were comparably distributed at the channel margins or at the centre (58%
and 42%, respectively, before and 53% and 47%,
respectively, after 2000). The surface area of each
habitat unit was measured (or visually estimated
when measurement was difficult) and depended on
the inherent variability of areas sampled without
enclosures (e.g. at the centre channel, the area sampled partly depended on the flow velocity). Habitat
unit depth was estimated as the average of three to 10
point measurements.
size-dependent. We analysed the variations of the
relative abundance of specific size classes to provide a
better understanding of community dynamics. For
this purpose, we defined two size classes for each of
the 10 most abundant species of the study reach in
order to separate individuals of the year and older
fish. Size class limits were defined for each species
and each survey using size histograms (limits ranged
between 5 and 10 cm). For consistency, only autumn
surveys were used for this analysis, as most young of
the year are easier to catch and identify in autumn, i.e.
at least 3 months after the main spawning period of
most species (March to June; Cattanéo et al., 2001).
Describing community changes
Community changes were assessed using species
richness and fish density (number of individuals per
square metre) and the relative abundance of species
(as percentage) calculated for each field survey (habitat units pooled). We tested if the means of these
variables differed before and after the minimum flow
change using t-tests. Principal components analysis
(PCA) of species relative abundances was used to
describe variations in community structure across
surveys. In PCA, relative abundances were ln(1 + x)
transformed to downweight the influence of a few
dominant species. We also tested if scores on the PCA
axes differed before and after the minimum flow
change using t-tests.
To test for the potential influence of rare but locally
abundant species on our results, we repeated the PCA
using the relative occurrence of species (percentage of
habitat units where the species occurred) in surveys
instead of the relative abundance. Finally, to test for
the influence of the habitat types sampled on our
results, we repeated the PCA on relative abundance
using separately data sampled at the channel margins
and data sampled at the channel centre.
Life stages of each species can be differently
affected by environmental changes. For example,
hydraulic changes modify simultaneously habitat
conditions for adults, availability of nutrients, near
bed spawning conditions and survival of fry. In
addition, community responses to environmental
changes are expected to occur over periods of several
years, at least longer than the lifespan of species
(around 5–6 years for fish of the Rhône river;
>15 years for, e.g. nase). Therefore, we expect community changes during the study to be dynamic and
Linking community structure with annual
environmental variables
Besides changes in minimum flow, annual variations
in community structure across surveys can be due to a
number of environmental characteristics (Grossman,
Dowd & Crawford, 1990; Poff et al., 1997). In particular, the frequency of high flows and temperature
patterns during key periods for fish population
dynamics can explain annual community variations.
This is particularly true in the context of the bypassed
reaches of the Rhône river, where high flows (that
contrast in magnitude with the minimum flow) and
temperature are quite variable among years. Similarly, discharge and temperature conditions preceding
field surveys may influence fish distributions during
the survey and thus capture efficiency. For example,
high flows preceding the survey may force fish to find
refuges along the banks where they are more easily
caught by electrofishing. Because of the paucity of
surveys available we limited the description of the
annual environment to seven variables describing low
flows during the year and high flows and temperature
during the reproduction period, the growth period
and the 15 days preceding the surveys (see Table 2 for
variable definitions). We tested the correlations
between annual community patterns (survey species
richness, survey fish density and coordinates on the
PCA axes) and these seven annual environmental
variables using autumn surveys (for consistency).
Testing predictions of fish community changes
A statistical instream habitat model was applied to
several bypassed reaches of the Rhône river and its
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
Fish community and minimum flow
main tributaries before the increase of minimum flow
at Pierre-Bénite (Lamouroux et al., 1999b,c). The
model was based on microhabitat preferences of the
14 more common species of the Rhône river. These
microhabitat preferences were derived from electrofishing data (habitat unit samples) collected using the
same methods as here but in other reaches of the basin
(in two large tributaries of the Rhône, the Ain and the
Ardèche, and in the bypassed reach of the Rhône at
Montélimar, 150 km downstream from Pierre-Bénite).
The model differed from conventional instream habitat model (e.g. Phabsim; Bovee, 1982) by the nature of
its hydraulic component (based on statistical approaches and using simple input data) and its biological
component (fish preference models were multivariate).
The statistical habitat model predicted two types of
attributes of fish communities from hydraulic conditions in reaches at minimum flow. First, relative
density indices (RI values) for the 14 species were
averages of the ln(1 + density) of species across all
habitat units sampled in the reach (all surveys pooled,
density being the number of individuals in 0.1 m3),
rescaled to sum to one across the 14 species considered here. As the relative proportions introduced
above, RI values reflected the expected weight of the
different species in the community. However, because
based on local ln(densities) their definition was more
complex. Ln(densities) are particularly interesting to
use because their variations are less sensitive to
random error in estimates of fish abundance or
habitat unit volume (see permutation tests in Lamouroux et al., 1999a). Second, two community indices
(CSI1 and CSI2) were defined as linear combinations
of RI values and summarised the predictions of
relative densities at the community level. Specifically,
predicted CSI1 and CSI2 were defined from a PCA on
predicted RI values in the Rhône catchment (i.e. for
different reaches pooled). CSI1 was strongly linked to
the longitudinal morphology of reaches; predicted to
be high when the RI values of riffle-dwelling species
were predicted to be high. CSI2 reflected the expected
effect of minimum discharge rate; predicted to be high
when the RI values of ‘midstream’ species preferring
fast-flowing and deep microhabitats were predicted to
be high.
Lamouroux et al. (1999c) showed that the observed
values of CSI1 and CSI2 of reaches (derived from
electrofishing data) were precisely predicted from
1735
hydraulics at minimum flow in 12 different reaches of
the Rhône river and its tributaries (including our
study reach at Pierre-Bénite before minimum flow
increase). Observed RI values were not well predicted
across sites because of zoogeographic differences
between the Upper Rhône (upstream from the city
of Lyon) and the Lower Rhône river. However,
observed differences in RI values between close sites
(i.e. pairs of close reaches with different hydraulic
conditions) were well predicted by the model. This
validation showed the potential of a regional instream
habitat model for predicting fish community variations across sites related to hydraulics. Based on these
results, Lamouroux et al. (1999b) simulated how
minimum flow changes at a site should modify RI
and CSI values. The minimum flow increase at PierreBénite is the first occasion for testing these predictions. The habitat model predicted an increase of CSI2
at Pierre-Bénite after minimum flow change, with
higher RI values for species preferring both fastflowing and/or deep habitat (barbel, bleak, nase, dace
and spirlin; see Table 1 for scientific names). CSI2 at
Pierre-Bénite after minimum flow increase should
have the highest value of all bypassed reaches of the
Rhône. By contrast, CSI1 values (and thus RI values of
riffle-dwelling species) were not expected to change
because they are essentially linked to the longitudinal
morphology of reaches (proportions of pools versus
riffles). Riffle-dwelling species should have contrasting responses to the increase in minimum flow,
depending on their preferences for water depth.
To test the predictions of Lamouroux et al. (1999b,c),
observed changes in RI values at Pierre-Bénite due to
minimum flow increase (differences between before
and after flow restoration) were compared with those
predicted from hydraulics at minimum flow (i.e. from
the statistical instream habitat model). Additionally,
we tested if the observed CSI2 value at Pierre-Bénite
after minimum flow increase reached the highest
value of all bypassed reaches of the Rhône river and if
the observed CSI1 value did not change, as predicted
by the habitat model.
Results
A strong annual variability in discharge patterns
among years was noted during the study. For example, 1999 and 2001 were ‘wet’ years with short periods
at the minimum flow, whereas the duration of
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
1736
N. Lamouroux et al.
Table 1 Species sampled at Pierre-Bénite ordered by their total abundance (1995–2004). The two last columns (% pre, % post) indicate
the relative abundance of species pre- and postrestoration of the minimum flow (per cent of individuals belonging to each species
averaged across surveys for each period)
Species code
Common name
Scientific name
Total abundance
% pre
% post
CHE
GAR
ABL
LOF
GOU
BAF
PES
BRB
HOT
SPI
CHA
PER
ROT
PSR
SIL
VAN
VAI
TAN
BOU
GRE
ANG
EPI
PCH
SAN
CAA
TAC
BLE
BRO
CCO
BRE
TOX
TRF
Chub
Roach
Bleak
Stone loach
Gudgeon
Barbel
Pumpkinseed
White bream
Nase
Spirlin
Sculpin
Perch
Rudd
Topmouth gudgeon
Wels
Dace
Minnow
Tench
Bitterling
Pope
Eel
Threespine stickleback
Catfish
Pike perch
Goldfish
Rainbow trout
River blenny
Pike
Common carp
Bream
Soiffe
Brown trout
Leuciscus cephalus (L., 1758)
Rutilus rutilus (L., 1758)
Alburnus alburnus (L., 1758)
Barbatula barbatula (L., 1758)
Gobio gobio (L., 1758)
Barbus barbus (L., 1758)
Lepomis gibbosus (L., 1758)
Blicca bjoerkna (L., 1758)
Chondostroma nasus (L., 1758)
Alburnoides bipunctatus (Bloch, 1782)
Cottus gobio (L., 1758)
Perca fluviatilis (L., 1758)
Scardinius erythrophthalmus (L., 1758)
Pseudorasbora parva (Temmick and Schlegel, 1842)
Silurus glanis (L., 1758)
Leuciscus leuciscus (L., 1758)
Phoxinus phoxinus (L., 1758)
Tinca tinca (L., 1758)
Rhodeus sericeus (Pallas, 1776)
Gymnocephalus cernuus (L., 1758)
Anguilla anguilla (L., 1758)
Gasterosteus aculeatus (L., 1758)
Ictalurus melas (Rafinesque, 1820)
Stizostedion lucioperca (L., 1758)
Carassius auratus (L., 1758)
Oncorhynchus mykiss (Walbaum, 1792)
Salaria fluviatilis (Asso, 1801)
Esox lucius (L., 1758)
Cyprinus carpio (L., 1758)
Abramis brama (L., 1758)
Chondostroma toxostoma (Vallot, 1837)
Salmo trutta (L., 1758)
2337
1857
1340
901
854
564
493
465
174
150
97
68
59
52
52
45
43
40
39
39
25
22
13
9
8
8
5
4
4
2
1
1
24.4
20.0
10.9
13.5
8.0
4.8
8.4
2.4
0.7
0.9
1.4
0.5
0.5
0.6
0.4
0.2
0.0
0.4
0.4
0.4
0.3
0.2
0.1
0.1
0.0
0.1
0.1
0.0
0.0
0.0
0.0
0.0
18.9
17.4
22.2*
2.7*
7.7
7.8
2.8
7.7*
3.2*
2.4*
0.7
1.3*
0.5
0.1
0.8
0.7
0.8
0.4
0.5
0.4
0.2
0.3
0.3
0.1
0.2
0.0
0.0
0.1
0.0
0.0
0.0
0.0
*Significantly different from % pre (t-test, P < 0.05).
minimum flow was high in ‘dry’ years, 1996 and 2003
(Fig. 3). Two floods of return period >10 years
occurred in March 2001 and November 2002, with a
maximum daily discharge >3400 m3 s)1 in the bypassed reach. Temperature patterns were also variable among years (Fig. 2), e.g. the summer 2003 was
particularly warm. However, water temperature did
not show any increasing trend during the study
period despite a context of global warming in the
region (Daufresne et al., 2003).
Describing community changes
A total of 9771 fish belonging to 32 species were
sampled during the study period. Neither survey
species richness (P ¼ 0.92) nor density (P ¼ 0.16)
differed before and after minimum flow increase (ttests). Survey richness varied between 16 (June 2002)
and 26 species (September 1998), indicating that all
species were not sampled during each survey. Survey
fish density varied between 0.16 (June 2002) and 1.61
individuals m)2 (September 2002).
The relative abundances of species reflect a predominance of a limited number of species both before
and after minimum flow increase (Table 1). Chub and
roach were the most abundant species during both
periods and 10 species represented 93% of the fish
captured (chub, roach, bleak, stone loach, gudgeon,
barbel, pumpkinseed, white bream, nase and spirlin).
However, the relative abundance of six of the 12 most
abundant species changed significantly after minimum flow increase (Table 1). Bleak, nase, spirlin,
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
Fish community and minimum flow
Table 2 Pearson’s correlation between
annual environmental variables and fish
community characteristics (survey species
richness, survey density and scores on the
first two axes of a PCA of relative abundance) for the eight autumn surveys. Bold
correlations indicate a significant association (P < 0.05).
1737
Environmental variable
Richness
Density
PCA-axis1
PCA-axis2
Low flows*
High flows – reproduction†
High flows – growth‡
High flows – survey§
Temperature – reproduction–
Temperature – growth**
Temperature – survey††
0.07
)0.34
)0.49
0.19
0.27
0.49
)0.08
0.26
)0.44
0.39
0.02
0.49
)0.05
0.15
0.93
)0.42
)0.14
)0.11
0.01
)0.23
)0.28
0.16
0.87
0.12
0.53
)0.46
)0.05
)0.38
*Number of days with discharge <25 m3 s)1 from the reproduction period to the
sampling period, i.e. between March and August.
†
Number of days with discharge >500 m3 s)1 for the reproduction period, i.e. March to
June.
‡
Idem for the growth period, i.e. July and August.
§
Idem for the 15 days preceding the survey.
–
Average daily temperature for the reproduction period – n ¼ 6 for this variable because
of missing temperature values.
**Idem for the growth period.
††
Idem for the 15 days preceding the survey. Limits of periods are from Cattanéo et al.
(2001).
18
Minimum
flow
change
16
% of barbel
14
12
10
8
6
4
2
0
01/1995 01/1997 01/1999 01/2001 01/2003 01/2005
Date
Fig. 4 Proportion of barbel (% of individuals) in the 12 field
surveys.
white bream and perch increased between two- and
4.3-fold, whilst stone loach decreased by a factor of
five. Although not significant individually, the relative abundance of barbel and dace were greater, while
sculpin, pumpkinseed and topmouth gudgeon were
lower after minimum flow increase. However, the
relative abundance of any species considered separately was highly variable among surveys (e.g. barbel,
Fig. 4).
The first PC axis explained 36% of the variance in
relative abundances and clearly discriminated surveys made pre- and postrestoration (t-test, P < 0.001;
Fig. 5), thereby supporting the changes in species
proportions described above. The 1999 survey
appeared as an exception, with a position on the
PCA close to postrestoration surveys. The second PC
axis, explaining another 18% of the variance in
relative abundances, did not discriminate pre- and
postrestoration surveys (t-test, P ¼ 0.46) but identified
surveys with relatively higher proportions of stone
loach and lower proportions of roach and pumpkinseed within both periods (e.g. October 1995, September 1999 and October 2001). Similarly, three additional
PCAs made on species occurrence, or using habitat
units sampled in one habitat type (channel margins or
centre) discriminated surveys pre- and postrestoration
on their first (t-tests, P ¼ 0.005, P ¼ 0.005, P ¼ 0.01,
respectively) but not on their second (t-tests, P ¼ 0.88,
P ¼ 0.13, P ¼ 0.72, respectively) axes.
Changes in relative abundance of the 10 predominant species at Pierre-Bénite were dynamic and sizedependent (Fig. 6). For species whose relative
abundance increased after minimum flow change
(i.e. bleak, nase, white bream, spirlin and barbel;
Table 1), this increase was essentially because of the
catch of numerous young of the year. Small fish of
these species had a higher cumulated relative abundance in and after 1999. The cumulated relative
abundance of large fish of these species tended to be
higher after 2001, but this effect is less clear (Fig. 6a).
By contrast, for the other species (chub, roach, stone
loach, pumpkinseed and gudgeon), the cumulated
relative abundance of large fish was lower in and after
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
1738
N. Lamouroux et al.
Fig. 6 Cumulated relative abundance of small (full diamonds)
and large (empty circles) fish of bleak, nase, white bream, spirlin
and barbel (a) and chub, roach, stone loach, pumpkinseed and
gudgeon (b) for eight field surveys (autumn surveys only).
Fig. 5 Plots of survey (a) and species (b) scores on the first (PC1)
and second (PC2) axes of a PCA of relative abundance. Surveys
pre- and postflow restoration are grouped to show their separation along PC1. Species codes are from Table 1. Species contributing little to the definition of axes were not represented for
readability.
correlated with low flows (R2 ¼ 0.86, P < 0.01), consistent with the separation of pre- and postrestoration
surveys on this axis (Fig. 5). Scores on the second PC
axis were correlated with high flows during the
reproduction period (R2 ¼ 0.76, P < 0.01), i.e. high
flows during the reproduction period favoured stone
loach compared with roach and pumpkinseed.
Testing predictions of fish community changes
1999. The cumulated relative abundance of small fish
of these species was lower in 2003 and 2004 (Fig. 6b).
Linking community structure with annual
environmental variables
Survey species richness and density in autumn were
not significantly correlated with any of the annual
environmental variables describing low or high flows
or temperature conditions (Table 2). Scores of autumn
surveys on the first PC axis were only significantly
Of the 14 species included in the habitat model of
Lamouroux et al. (1999c), 13 were present at PierreBénite, representing 91% of the fish captured during
our study. For these 13 species the observed changes
in RI values between the pre- and postrestoration
periods were significantly related to those independently predicted from hydraulics by the habitat model
(R2 ¼ 0.42, P ¼ 0.02, slope not different from 1;
Fig. 7). RI values for barbel (BAF), bleak (ABL), nase
(HOT) and spirlin (SPI) increased as expected or
more; RI values for gudgeon (GOU) decreased less
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
Fish community and minimum flow
10
0.18
ABL
BAF
ce
5
SPI
GAR
VAI PER
0
VAN
ANG
GOU
CHE
–5
LOF
y=
ard
pb1
y
0
5
=
x
pr
10
–0.18
–0.18
Predicted change in RI
Ice
mo
Ibc bc Iby
PES
–5
by
pb2
0.00
–0.09
x
–10
–10
ain
0.09
HOT
Observed CSI1
Observed change in RI
1739
–0.09
0.00
0.09
0.18
Predicted CSI1
Fig. 7 Change in relative density indices of species (RI) at
Pierre-Bénite between the pre- and postflow restoration period:
observed change (electrofishing) versus predicted change (habitat model; Lamouroux et al., 1999a,b) and associated regression
line (y ¼ 0.89x + 0.12, R2 ¼ 0.42, P ¼ 0.02).
0.12
pb2
Discussion
Our study provides the first analysis of a minimum
flow increase below a dam in a large river with many
species. Only 4 years after flow restoration at PierreBénite (i.e. August 2000), several of our results
indicate a significant, predicted change in fish community structure related to minimum flow enhancement. First, surveys made before and after minimum
flow increase were clearly separated along the first
axis of a PCA of specific relative abundances. The
1999 survey appeared as an exception in this analysis,
with a community structure close to that observed
after minimum flow change. However, 1999 was
hydrologically an unusual year, with very few days
at low flows (n ¼ 38) compared with the previous
years (n > 71). Second, a change in community struc-
Observed CSI2
0.06
than expected and RI values for pumpkinseed (PES)
and stone loach (LOF) decreased more than expected
after minimum flow change.
The two synthetic indices related to the proportion
of riffle-dwelling (CSI1) and ‘midstream’ species
(CSI2) changed as predicted by the habitat model at
Pierre-Bénite after minimum flow increase (Fig. 8).
CSI1 did not change after minimum flow increase and
was low compared with many other reaches of the
Rhône and its tributaries. By contrast, CSI2 increased
and reached the highest value observed in bypassed
reaches of the Rhône river.
by
Iby
0.00
ard
pb1
ce
Ice
mo
bc
ain
Ibc
–0.06
y
–0.12
–0.12
=
x
pr
–0.06
0.00
0.06
Predicted CSI2
0.12
Fig. 8 Values of community structure indices (CSI1 and CSI2) at
Pierre-Bénite before (pb1) and after (pb2) minimum flow increase: observed values (electrofishing) versus predicted values
(habitat model; Lamouroux et al., 1999a,b). The position of other
bypassed reaches of the Rhône river (mo: Montélimar; pr: Péagede-Roussillon; bc and Ibc: two reaches at Brégnier-Cordon, Ibc
being influenced by a weir; by and Iby: two reaches at Belley; ce
and Ice: two reaches at Chautagne) and of main tributaries of the
Rhône (ain: Ain; ard: Ardèche) are added for a comparison (data
from Lamouroux et al., 1999b). Note that the position of pb1 is
(very) slightly different from the position of Pierre-Bénite before
restoration shown by Lamouroux et al. (1999b) because of the
availability of additional fish data here.
ture across autumn surveys (PCA axis 1 scores; Fig. 5)
was significantly related to the number of days at low
flows in spring and summer, but not to any other
annual environmental variable describing high flows
and annual temperature at key periods of fish life
cycle. Third, the observed change in community
structure was observed for different treatments of
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
1740
N. Lamouroux et al.
our data (analyses of the relative abundance of
individual species, PCA of relative abundance and
occurrences, separate analyses of data collected along
the margins or at the channel centre). Finally, changes
in specific density and community structure indices
observed at Pierre-Bénite match those predicted from
an independent habitat model and are consistent with
the knowledge of species microhabitat preferences.
Most species whose relative abundance increased at
Pierre-Bénite prefer deep and/or fast-flowing hydraulic conditions. For example, bleak is a pelagic species
(Copp, 1992) preferring intermediate velocities
(Lamouroux et al., 1999a; Bruslé & Quignard, 2001).
Both nase, which migrated to France from central
Europe via artificial channels at the end of the 19th
century and barbel are rheophilic species typical of
large, fast-flowing rivers (Nelva, 1988; Bruslé &
Quignard, 2001). Spirlin, also a rheophilic species,
prefers oxygenated habitats with intermediate depths
(Lamouroux et al., 1999a; Bruslé & Quignard, 2001).
Nase, barbel and spirlin are also lithophilic species
that spawn on gravel or cobble in fast-flowing
habitats. The increase of the relative abundance of
white bream and perch were less expected because
these species generally avoid fast-flowing habitats.
Perch, however, is not abundant at Pierre-Bénite and
white bream was often caught in macrophyte patches
where it spawns and where velocity is locally
reduced. Overall, the minimum flow increase at
Pierre-Bénite favoured fluvial species that have been
largely affected by the regulation of the large rivers in
the Rhône (Persat, 1988; Lamouroux et al., 1999c) and
elsewhere (Schiemer & Spindler, 1989). Among the
species whose relative abundance decreased, we note
both species avoiding fast flows (pumpkinseed and
topmouth gudgeon) and small rheophilic species that
avoid deep water (stone loach and sculpin) (Bruslé &
Quignard, 2001).
Estimating fish abundances in large rivers is
difficult, often resulting in high levels of uncertainty
(Persat & Copp, 1990). A fish survey at Pierre-Bénite
covered only 0.2% of the total surface area of the
reach and as many stream fish are notoriously
gregarious (Bruslé & Quignard, 2001) their capture
in our habitat units is variable. In this context, our
quantitative estimates of a particular species proportion should be considered as potentially biased. It is
only the number of species affected by flow restoration, the consistency of observed changes for different
treatments of the data and the predictability of these
changes that enable us to conclude on a significant
change in community structure. It is not surprising
that fish community changes were essentially
observed for community variables depending on the
relative abundance of species (Daufresne et al., 2003),
as these type of variables are better estimated by our
sampling design than other fish community descriptors (richness and density). Estimates of survey
richness and density were probably affected by the
limits of fish sampling in large rivers. The difference
of fish density estimated for example in June 2002
(0.16 individuals m)2) and 3 months after in September 2002 (1.61 individuals m)2) illustrates these difficulties. Moreover, survey species richness was always
lower than the total species richness of the reach (32,
all surveys pooled) and its variations were because of
the uncertain capture of a number of rare species (e.g.
trout, pike perch, bream and river blenny; see
Table 1). A study of these community descriptors
would require more intensive sampling effort in space
and time (see also Lamouroux et al., 1999c) and our
difficulty to interpret their temporal variation is an
obvious limit of our study.
Our study suggests a very rapid change in species
relative abundance that was observed since 2001 and
even during the ‘wet’ year 1999. The rapidity of the
fish response was unexpected, because most fish
species at Pierre-Bénite have life cycles of 5 years or
more (Bruslé & Quignard, 2001). Furthermore, former
observations of fish community response to environmental changes suggested responses over a longer
period (e.g. Matthews, Schorr & Meador, 1996; Eklöv
et al., 1998; Pegg & McClelland, 2004). Our analysis of
size class responses provides a rational explanation
for this rapid change. Species whose relative abundance increased at Pierre-Bénite may have benefited
from an immediate higher reproduction success,
because it is essentially the relative proportion of
small fish of these species that increased. The increase
of relative abundance of large individuals of these
species is less evident and was not immediate after
minimum flow change. Minimum flow increase may
have enhanced the usable area and the spawning
conditions (generally fast-flows on coarse substrate) of
species preferring deep and fast-flowing hydraulic
habitats. By contrast, it is essentially the large
individuals of species whose relative abundance
decreased that are responsible for the rapidity of the
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
Fish community and minimum flow
observed community change (Fig. 6). This suggests
that adults of these species either had higher mortalities after minimum flow change or escaped from the
reach to find more suitable hydraulic conditions (e.g.
downstream of our study reach in Vaugris reservoir).
The modification of hydraulic conditions in dike fields
along the shoreline, following minimum flow
increase, may also have influenced the survival of
young of the year (Nicolas & Pont, 1997). Overall, the
different responses of specific size classes and the
high variability of community patterns observed
across surveys after minimum flow change reveal
that the fish community at Pierre-Bénite is in a
transitional phase. Continued monitoring of the fish
life stages will indicate if the changes observed at
Pierre-Bénite are perennial or not.
In addition to the (transitional) change in community structure observed after minimum flow increase,
the annual variability of the environment may explain
a minor part of the variability in community structure
across surveys. In particular, we found a significant
correlation between the frequency of high flows
during the reproduction period and the second PC
axis of relative abundance. High flows (and maybe
their indirect action on substrate) favoured the reproduction of stone loach, a species spawning in clear
riffles (Bruslé & Quignard, 2001), despite the general
decrease of this species because of minimum flow
increase. A more intensive sampling design, including
regular sampling from March to September, would
probably allow a more precise interpretation of the
observed inter-annual variability.
The case-study of Pierre-Bénite supports the idea
that fish community responses to hydraulics are
partly predictable in large rivers. It expands the scope
of the few studies indicating an impact of minimum
flow increase on fish populations in smaller streams
(Phillips et al., 2000; Jowett & Biggs, 2004). Large
rivers are inhabited by species that are adapted to
deep, fast-flowing systems (Persat, 1988; Schiemer &
Spindler, 1989; Penczak & Kruk, 2000; Galat &
Zweimüller, 2001), i.e. the main channel of large
rivers cannot be viewed as ‘highways’ where fish only
transit (Junk, Bailey & Sparks, 1989). Moreover, the
results obtained at Pierre-Bénite match general expectations derived from the application of instream
habitat models in different countries (Lamouroux &
Jowett, 2005). Indeed, habitat values derived from
instream habitat models are sensitive to two hydraulic
1741
axes in many rivers: a geomorphological axis (corresponding to our index CSI1) where riffle-dwelling
species contrast with pool-dwelling ones, and a
discharge rate axis (corresponding to our CSI2) where
midstream species adapted to fast-flowing and deep
microhabitats contrast with others (Lamouroux &
Jowett, 2005). Correspondingly, rheophilic species had
contrasted response to minimum flow enhancement at
Pierre-Bénite. Rheophilic, midstream species preferring intermediate or high water depth were favoured
by the minimum flow increase at Pierre-Bénite
relative to other rheophilic, riffle-dwelling species
preferring shallow conditions (stone loach, sculpin).
Rheophilic species as a whole are often used in
biomonitoring for indicating the effects of physical
modifications of rivers (Oberdorff et al., 2002), or in
general definitions of habitat guilds (e.g. Aarts et al.,
2004). Our case study suggests the need to refine guild
definitions in such studies, by separating rheophilic
species sensitive to discharge management from
species sensitive to morphological changes of the
streambed (i.e. weir addition or removal, management of the bed morphology). Pierre-Bénite is the first
of six reaches of the Rhône river where minimum flow
has been or should be increased. We expect this case
study to be a base for a quantitative comparison of
other experiences of flow restoration in large rivers, in
the Rhône catchment and elsewhere.
Acknowledgments
Financial support was provided by the Compagnie
Nationale du Rhône (CNR), the Agence de l’Eau
Rhône-Méditerranée-Corse, the Diren Rhône-Alpes,
and the Syndicat mixte du Rhône des ı̂les et des lônes
(SMIRIL). Sylvie Valentin coordinated the fish surveys
before 1999. Georges Carrel, Richard Johnson, Michael
Ovidio, Yves Souchon and anonymous reviewers
provided helpful comments.
References
Aarts B.G.W., Van den Brink F.W.B. & Nienhuis P.H.
(2004) Habitat loss as the main cause of slow recovery
of fish faunas of regulated large rivers in Europe: The
transversal floodplain gradient. River Research and
Applications, 20, 3–23.
Amoros C. (2001) The concept of habitat diversity
between and within ecosystems applied to river side-
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
1742
N. Lamouroux et al.
arm restoration. Environmental Management, 28, 805–
817.
André S. & Olivier J.M. (2003) Mise en place du suivi
scientifique du programme décennal de restauration hydraulique et écologique du fleuve Rhône – rapport d’étape du
deuxième comité de pilotage convention DIREN-EZUS
Lyon 1. Technical Report, Direction Régionale de
l’Environnement Rhône-Alpes, Lyon, France.
Bovee K.D. (1982) A Guide to Stream Habitat Analysis
Using the Instream Flow Incremental Methodology. Instream Flow Information Paper 12, U.S. Fish and Wildlife
Service, Fort Collins, CO.
Bravard J.P., (1987) Le Rhône du Léman à Lyon. La
Manufacture, Lyon, France.
Bruslé J. & Quignard J.-P. (2001) Biologie des poissons d’eau
douce européens. Technique et Documentation, Paris,
France.
Cattanéo F., Carrel G., Lamouroux N. & Breil P. (2001)
Relationships between hydrology and cyprinid reproductive success in the lower Rhône at Montélimar,
France. Archiv für Hydrobiologie, 151, 427–450.
Copp G.H. (1992) Comparative microhabitat use of
cyprinid larvae and juveniles in a lotic floodplain
channel. Environmental Biology of Fishes, 33, 181–193.
Daufresne M., Roger M.C., Capra H. & Lamouroux N.
(2003) Long-term changes within the invertebrate and
fish communities of the Upper Rhône River: effects of
climatic factors. Global Change Biology, 10, 124–140
Eklöv A.G., Greenberg L.A., Brönmark C., Larsson P. &
Berglund O. (1998) Response of stream fish to
improved water quality: a comparison between the
1960s and 1990s. Freshwater Biology, 40, 771–782.
Fruget J.F. (1992) Ecology of the lower Rhône after 200 years
of human influence: a review. 7, 233–246.
Galat D.L. & Zweimüller I. (2001) Conserving large-river
fishes: is the highway analogy an appropriate paradigm? Journal of the North American Benthological
Society, 20, 266–279.
Gore J.A. & Nestler J.M. (1988) Instream flow studies in
perspective. Regulated Rivers. Research and Management, 2, 93–101.
Grossman G.D., Dowd J.F. & Crawford M. (1990)
Assemblage stability in stream fishes. Environmental
Management, 14, 661–671.
Harris D.D., Hubert W.A. & Wesche T.A. (1991) Brown
trout population and habitat response to enhanced
minimum flow in Douglas Creek, Wyoming. Rivers, 2,
285–294.
Jowett I.G. & Biggs B.J.F. (2004) The effectiveness of
habitat-based minimum flow assessments - six case
studies. In: Proceedings of the Fifth International Symposium on Ecohydraulics (Eds D. Garcia de Jalon & P.
Vizcaino Martinez), pp. 695–701. IAHR, Madrid, Spain.
Junk W.J., Bailey P.B. & Sparks R.E. (1989) The flood
pulse concept in river-floodplain systems. In: Proceedings of the International Large River Symposium (Ed. D.P.
Dodge), pp. 110–127. Canadian Special Publication of
Fisheries and Aquatic Sciences, Ottawa, 106.
Lamouroux N. & Jowett I.G. (2005) Generalized instream
habitat models. Canadian Journal of Fisheries and Aquatic
Sciences, 62, 7–14.
Lamouroux N., Capra H., Pouilly M. & Souchon Y.
(1999a) Fish habitat preferences at the local scale in
large streams of southern France. Freshwater Biology,
42, 673–687.
Lamouroux N., Doutriaux E., Terrier C. & Zylberblat M.
(1999b) Modélisation des impacts de la gestion des
débits réservés du Rhône sur les peuplements piscicoles.
Bulletin Français de la Pêche et de la Pisciculture, 352, 45–61.
Lamouroux N., Olivier J.M., Persat H., Pouilly M.,
Souchon Y. & Statzner B. (1999c) Predicting community characteristics from habitat conditions: fluvial fish
and hydraulics. Freshwater Biology, 42, 1–25.
Lamouroux N., Chandesris A., Fruget J.F. et al. (2004)
Chute de Pierre-Bénite. Suivi de l’incidence de l’augmentation du débit réservé dans le vieux-Rhône. Phase II (2001–
2004). Technical Report, Cemagref-Lyon & Compagnie
Nationale du Rhône, Lyon, France.
Matthews W.J., Schorr M.S. & Meador M.R. (1996) Effects
of experimentally enhanced flows on fishes of a small
Texas (U.S.A.) stream: assessing the impact of interbasin transfer. Freshwater Biology, 35, 349–362.
Nelva A. (1988) Profil écologique du hotu, Chondrostoma nasus (Ostéichtyen, Cyprinidés) dans le HautRhône français: structure d’habitat ponctuel et stationnel. Acta Oecologia, 9, 275–296.
Nicolas Y. & Pont D. (1997) Hydrosedimentary Classification of Natural and Engineered Backwaters of a Large River,
the Lower Rhone: Possible Applications for the Maintenance
of High Fish Biodiversity. Regulated Rivers: Research
and Management, 13, 417–431.
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.
Pegg M.A. & McClelland M.A. (2004) Spatial and
temporal patterns in fish communities along the
Illinois river. Ecology of Freshwater Fish, 13, 125–135.
Penczak T. & Kruk A. (2000) Threatened obligatory
riverine fishes in human-modified Polish rivers.
Ecology of Freshwater Fish, 9, 109–117.
Persat H. (1988) De la biologie des populations de
l’ombre commun Thymallus thymallus (L.1758) à la
dynamique des communautés dans un hydrosystème
fluvial aménagé, le Haut-Rhône français. State Thesis,
Université Lyon 1, Villeurbanne, France.
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743
Fish community and minimum flow
Persat H. & Copp G.H. (1990) Electric fishing and point
abundance sampling for the ichthyology of large
rivers. In: Developments in Electric Fishing (Ed. I.G.
Cowx), pp. 197–209. Fishing News Books, Cambridge
University Press, Cambridge, UK.
Phillips J.L., Ory J. & Talbot A. (2000) Anadromous
salmonid recovery in the Umatilla river basin, Oregon:
a case study. Journal of the American Water Resources
Association, 36, 1287–1308
Poff N.L., Allan J.D., Bain M.B., Karr J.R., Prestegaard
K.L., Richter B.D., Sparks R.E. & Stromberg J.C. (1997)
The Natural Flow Regime. A paradigm for river
conservation and restoration. BioScience, 47, 769–784.
Poff N.L., Allan J.D., Palmer M.A., Hart D.D., Richter
B.D., Arthington A.H., Rogers K.H., Meyer J.L. &
Stanford J.A. (2003) River flows and water wars:
emerging science for environmental decision making.
Frontiers in Ecology and the Environment, 1, 298–306.
Propst D.L. & Gido K.B. (2004) Responses of native and
nonnative fishes to natural flow regime mimicry in the
1743
San Juan river. Transactions of the American Fisheries
Society, 133, 922–931.
Richter B.D., Mathews R., Harrison D.L. & Wigington R.
(2003) Ecologically sustainable water management:
managing river flows for ecological integrity. Ecological
Applications, 13, 206–224.
Schiemer F. & Spindler T. (1989) Endangered Fish Species of
the Danube River in Austria. 4, 397–407.
Souchon Y. & Capra H. (2004) Aquatic habitat modelling:
biological validations of IFIM/Phabsim methodology
and new perspectives. Hydroécologie Appliquée, 14, 9–25.
Thévenet A. & Statzner B. (1999) Linking fluvial fish
community to physical habitat in large woody debris:
sampling effort, accuracy and precision. Archiv für
Hydrobiologie, 145, 57–77.
Vadas R.L. Jr & Orth D.J. (1993) A new technique for
estimating the abundance and habitat use of stream
fishes. Journal of Freshwater Ecology, 8, 305–317.
(Manuscript accepted 15 June 2006)
Ó 2006 The Authors, Journal compilation Ó 2006 Blackwell Publishing Ltd, Freshwater Biology, 51, 1730–1743