Fisheries Research 96 (2009) 70–76

Contents lists available at ScienceDirect

Fisheries Research
journal homepage: www.elsevier.com/locate/fishres

Assessing fish assemblages in reed habitats of a large shallow

lake—A comparison between gillnetting and electric fishing
T. Erős ∗ , A. Specziár, P. Bíró
Balaton Limnological Research Institute of the Hungarian Academy of Sciences, Klebelsberg K. u. 3., H-8237 Tihany, Hungary

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

Keywords: To know the possible bias of different fishing methods is essential in fisheries management, ecology and
Shallow lakes conservation. In this study species number, abundance, biomass, length class distributions and predefined
Fish communities ecological features of fishes were compared across two data sets collected simultaneously using gillnet
Multi-mesh gillnets
sampling and electric fishing in the reed habitats of the shallow and eutrophic Lake Balaton, Hungary.
Electric fishing
Sampling effort
With increasing sample size, electric fishing proved to be more effective in detecting new species, and
samples collected with this method were more species rich when standardized to the number of individ-
uals collected. Ordinations based on relative abundance and biomass data indicated highly contrasting
differences between the two methods. Bleak were caught by multi-mesh gillnets in much higher relative
numbers. However, the shape and size selectivity of the gillnets also reinforced differences between the
two methods. Size distribution data showed that gillnets caught relatively more middle-sized fish com-
pared with electric fishing. Estimates of the abundance and biomass of non-native species by gillnetting
and electric fishing differed, and differences were found in the proportions of various guilds (feeding,
spawning and habitat). However, it was not possible to conclude which gear’s estimate is closer to reality.
The study illustrates that reliance on single-gear surveys can be misleading in assessing fish assemblages
in reed habitats of a large and shallow, eutrophic lake.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction
The accurate estimation of biotic assemblage attributes (e.g.
species richness and composition, relative abundances, functional
metrics) is a fundamental requirement in environmental monitor-
ing and assessment (Cao et al., 2003; Kennard et al., 2006). For
sampling fish assemblages a variety of catching methods are avail-
able (Cowx, 1996; Murphy and Willis, 1996). However, catching
effectiveness, including species selection and size selectivity pat-
terns may differ for each gear, making it difficult to determine
whether these differences allow for accurate characterization of
assemblage attributes. Hence, there is a need for a more intensive
evaluation of between-gear variations to determine their relative
efficiency (Casselman et al., 1990; Jackson and Harvey, 1997; Cowx
et al., 2001; Olin and Malinen, 2003; Goffaux et al., 2005).
Fish assemblage metrics are being intensively used to determine
the ecological status of lakes (Randall and Minns, 2002; Gassner et
al., 2003; Drake and Valley, 2005; Garcia et al., 2006), and fish as a
group is one of the key biotic elements for such evaluation under
the Water Framework Directive of the European Union (EU, 2000).
∗ Corresponding author. Tel.: +36 87 448 244; fax: +36 87 448 006.
E-mail address: ertib@tres.blki.hu (T. Erős).
While there is a great need for a standardized fishing protocol for
lake habitats, such a methodology is still under development in
European countries. The two recommended sampling methods that
are used most often are gillnet sampling (Appelberg et al., 1995;
CEN, 2005) and electric fishing (CEN, 2003). Gillnets are passive
tools, and as such their catching effectiveness depends largely upon
fish activity. Even if multi-mesh gillnets are used, their species, mor-
phological and size selectivity can be significant. Electric fishing, an
active method, is generally considered to be the most adequate sin-
gle tool for sampling shallow (i.e. less than 1.5 m deep) freshwater
habitats (see e.g. Cowx, 1996). However, the efficiency of electric
fishing can vary significantly with the physical and chemical char-
acteristics of the habitat, the device used, etc. As a consequence,
this sampling method can also show important species and size
selectivity.
Very few studies have examined comparatively the effectiveness
of gillnetting and electric fishing when assessing the assemblage
attributes of fishes either from rivers (Growns et al., 1996; Goffaux
et al., 2005) or lakes (Vaux et al., 2000). Vaux et al. (2000) concluded
that when developing Environmental Monitoring and Assessment
protocols the number of fish species and individuals collected per
4-min transect by electric fishing (Coffelt Mark-10 350 w backpack
gear) was either superior or similar to that of gillnets set overnight
in the lakes of the northeastern United States. However, their gill-

0165-7836/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.fishres.2008.09.009
 T. Erős et al. / Fisheries Research 96 (2009) 70–76
netting procedure included samples from a diverse array of large
lake habitats (such as bottom sets in the hypolimnion, metalimnion,
and epilimnion and pelagic epilimnetic sets), while electric fish-
ing samples were confined to the littoral zone. Currently there are
no detailed comparative data on electric fishing and gillnetting for
samples taken from the same (or equivalent) lake habitats.
In studies of fish assemblages, generally, the open habitats of
lakes are sampled with gillnetting, while electric fishing is used in
the littoral zone along the shoreline or among macrophytes (e.g.
Diekmann et al., 2005; Mehner et al., 2005). In previous stud-
ies, therefore, fish assemblage comparisons between habitats also
implied differences in the bias of the sampling methods. Since the
relative bias of both electric fishing and gillnet sampling is largely
unknown, the aim of this study was to compare and evaluate elec-
tric fishing and gillnet catches, collected in the natural habitats of a
large Central-European lake. Reed habitat is one of the most impor-
tant natural habitat types in shallow eutrophic lakes and wetlands,
providing feeding, spawning and refuge areas for fish. Conse-
quently, we examined relationships between species richness and
sampling effort, assemblage composition, size class variations and
some important functional metrics (as used in fish-based evalu-
ations) to get a deeper insight into the relative effectiveness of
electric fishing and gillnetting in assessing fish assemblages in the
reed habitats of Lake Balaton, Hungary.
2. Methods
2.1. Study area and sampling procedure
Lake Balaton is one of the largest shallow lakes in Central Europe
(596 km2 ). Its mean depth is 3.2 m and the lake is meso-eutrophic
(Istvánovics et al., 2007). The littoral zone of the lake is mostly mod-
ified. Although widely distributed in the past, reed vegetated areas
are now fragmented, with a total area of only 11 km2 .
Electric fishing and gillnetting were carried out in the summer
of 2007 on five occasions (06, 07, 12, 27 June and 04 July). Five dis-
tinct reed habitats were examined, one each day, along the northern
shoreline of the lake. The sites belonged to the littoral zone and
were large enough to allow both electric fishing and gillnetting
with many replicate samples (see below). The reed was fragmented
enough for both setting the nets and conducting boat electric fish-
ing. Finally, the sites were remote enough from boat harbours to be
relatively infrequently used by anglers.
Electric fishing was carried out during daytime using a
generator-powered machine (Hans-Grassl Gmbh EL64 II GI; DC,
7.5 kW, 300–600 V; see http://www.hans-grassl.com/ accessed 17
March 2008). The cathode, a 5 m long copper cable, was floated at
the rear of the boat. To allow effective manuovering in the reed, a
small rubber boat (Yamaha 300S) with an electric engine was used.
The crew comprised two persons: one for catching the fish with
the hand-held anode (2.5 m long pole with a net of 40 cm diam-
eter, mesh size 6 mm) and one for driving the boat. Continuous
electrofishing was carried out by dipping the anode into the water
at approximately 3 m long intervals and pulling the anode toward
the boat, while moving slowly ahead. For each electric fishing 35 m
long sections were sampled (this being equal to the length of one
gillnet set). A high precision GPS device (Garmin GPSMAP 76Cx;
precision 2–4 m in the field) was used to measure the length of the
sampling route. Altogether 17, 19, 16, 18, and 17 replicates of 35 m
long sections were electrofished in the five habitats, which yielded
a total of 87 samples (3045 m). Fishing time for each section took
∼4 min. The captured fish were identified, their standard length
measured (mm) and then released at the end of each section. To
avoid recapture, each electric fishing started c. 10–15 m away from
the previously sampled section. Based on both the electric fish-
71
ing and the gillnet data (see below), NPUE was calculated as the
standardized total number of fishes collected per hour.
Similarly to electric fishing, gillnetting was carried out during
the daytime. Although it is suggested that gillnet sampling should
be carried out at night (Appelberg, 2000; CEN, 2005), deviation
from the standard was necessary for two reasons. First, setting gill-
nets in the daytime allowed comparison of the two methods during
the same time period. Second, daytime sampling avoided the rapid
accumulation of fish in the gillnets (see Olin et al., 2004), which
can happen in under one hour at night in Lake Balaton (Specziár,
2001). For gillnetting, 1.5 m high European standard benthic gill-
nets (Appelberg, 2000; CEN, 2005) rigged with two more 2.5 m long
panels of 65 and 80 mm mesh sizes were used. Thus our nets com-
prised 14 panels of 43, 19.5, 6.25, 10, 55, 8, 12.5, 24, 15.5, 5, 35, 29,
65 and 80 mm meshes and in total were 35 m long. The gillnets cov-
ered the whole water column in all cases. The duration of set was
between one and two hours during the period 09:00–11:00 a.m.
Such a relatively short time period was necessary to avoid excessive
accumulation of fish in the nets so as not to exceed the 6 kg catch
limit per net (CEN, 2005). At each site, five nets were set in random
locations, and consequently 25 nets were used for this study.
The European standard multi-mesh gillnet is known to be
ineffective for catching fish <5 cm (Appelberg, 2000; CEN, 2005).
Similarly, general electric fishing techniques are largely ineffective
for fish <2 cm, to an unknown extent (except devices specialized
for catching 0-group fish; see e.g. Cowx et al., 2001). To avoid bias
in our comparisons, all young-of-the-year fish were omitted from
the analyses. This age group (i.e. 0+) could be distinguished easily
by length-frequency analysis.
For biomass data, the weight of each individual fish was esti-
mated based on previous length-weight regressions, calculated
for the fishes of Lake Balaton (Specziár et al., 1997; Specziár
unpublished data). Mean water depth ± S.D. was 1.18 ± 0.23 m and
1.27 ± 0.24 m for electric fishing (n = 87) and gillnetting (n = 25)
samples, respectively, and did not differ between the sampling loca-
tions examined with the two methods (t-test, P < 0.001). Water
transparency (Secchi depth) was 0.62 ± 0.20 and 0.63 ± 0.23 m,
respectively, and did not differ between the sampling sites either
(t-test, P < 0.001). Conductivity, an important parameter for the effi-
ciency of electric fishing, varies between 550 and 671 ␮S/cm in Lake
Balaton (Specziár and Vörös, 2001).
2.2. Data analysis
Sample-based and individual-based rarefaction analyses were
used to examine changes in the estimated number of species as
a function of both number of samples and number of individuals
collected (Gotelli and Colwell, 2001; Colwell, 2005). Although it
is not easy to standardize samples collected with an active (elec-
tric fishing) and a passive (gillnetting) catching method, rarefaction
analyses can provide the least-biased evaluation of differences in
the number of species collected with the two methods.
Two-dimensional nonmetric multidimensional scaling (NMDS)
ordinations were run to visualise the variability and differences in
fish assemblage composition between the two sampling methods.
The analyses were run based on square-root-arcsine transformed
relative abundance and biomass data. To obtain robust and read-
ily interpretable results, samples from a certain reed site were
pooled according to the collection method applied. Consequently,
the positions of 10 sampling points (two collecting methods and
five spatially separated reed habitats) were compared in the anal-
yses. Species with an overall relative abundance or biomass of less
than 1% were combined into a single “rare species” group to pre-
vent the number of variables (i.e. species) from highly exceeding the
number of objects (sampling points). Since the relative abundance
72 T. Erős et al. / Fisheries Research 96 (2009) 70–76

Table 1
The species composition, relative abundance and biomass of fishes collected with gillnetting (N = 25 nets; 35 m each) and electric fishing (N = 87 × 35 m long sections; 3045 m
total)

Common name Scientific name Rel. abundance (%) Rel. biomass (%) Status Trophic Reproductive Habitat

Gillnet Electric. Gillnet Electric. Guild Guild Guild

Asp Aspius aspius 0.023 0.532 0.090 3.897 na Pisc Lith Non-b
Bitterling Rhodeus sericeus 0.129 3.727 0.021 0.064 na Insv/Herb Ostr Non-b
Bleak Alburnus alburnus 91.765 62.513 40.592 2.681 na Plankt Phli Non-b
Bream Abramis brama 0.340 0.639 3.732 1.129 na Omni Phli Bent
Carp Cyprinus carpio 0.070 0.852 9.905 34.024 na Omni Phyt Bent
Eel Anguilla anguilla – 5.112 – 23.416 nn Insv/Pisc ? Bent
European catfish Silurus glanis – 0.213 – 0.631 na Pisc Phyt Bent
Gibel Carassius gibelio – 0.532 – 2.666 nn Omni Phyt Non-b
Grass carp Ctenopharyngodon idella – 0.106 – 6.267 nn Herb Pel Non-b
Perch Perca fluviatilis 0.633 0.213 3.259 0.061 na Insv/Pisc Phli Non-b
Pike Esox lucius – 0.319 – 2.569 na Pisc Phyt Non-b
Pikeperch Sander lucioperca 0.035 1.278 0.004 12.247 na Pisc Poly Non-b
Pumpkinseed Lepomis gibbosus 0.023 1.065 0.029 0.082 nn Insv Psam Non-b
Roach Rutilus rutilus 4.030 10.863 26.758 5.430 na Omni Phli Non-b
Rudd Scardinius erythrophthalmus 0.340 10.330 3.476 3.657 na Omni Phyt Non-b
Ruffe Gymnocephalus cernuus 0.035 – 0.026 – na Insv Lith Bent
Tench Tinca tinca – 0.106 – 0.314 na Omni Phyt Bent
Volga pikeperch Sander volgensis 0.035 – 0.398 – na Pisc Poly Bent
White bream Blicca bjoerkna 2.518 1.597 11.707 0.865 na Omni Phli Bent
Whitefin gudgeon Romanogobio albipinnatus 0.023 – 0.003 – na Insv Psam Bent
Total number or biomass (kg) of fish 8537 939 106.21 126.8

Total number and biomass of fish collected for both techniques are also indicated at the bottom of the table. ‘status’ refers to whether the species is native (na) or non-native
(nn) to Lake Balaton. The abbreviations for trophic guilds are as follows: Herb, herbivores; Insv, insectivores; Plankt, planktivores; Omni, omnivores; Pisc, piscivores. The
abbreviations for reproductive guilds are as follows: Lith, lithophils; Ostr, ostracophils; Phyt, phytophils; Phli, phyto-lithophils; Pel, pelagophils; Psam, psammophils; Poly,
polyphils (non-specialist). Habitat guild refers to whether the species is benthic (Bent) or non-benthic (Non-b).

of bleak Alburnus alburnus comprised 88.9% of the total, two analy-
ses were performed for relative abundance data: one including all
fish species and one in which bleak was omitted from the analysis.
The Bray and Curtis dissimilarity coefficient was used as a distance
measure in all analyses. All two-dimensional analyses yielded a
stress value (the measure of the match between distances in the
original matrix and distances in the reduced ordination space) of
less than 5.3 and consequently proved to be highly interpretable. In
the program used, stress varies from 0 to 100, with low stress val-
ues (i.e. <20) indicating highly effective variance reduction (PC-Ord
Version 3.0, McCune and Mefford, 1997).
Between-gear differences in length distributions for all species
combined (i.e. the total sample) and for the two most abundant
fish (bleak and roach Rutilus rutilus) were compared using the Chi-
square test for independent samples.
Finally, Mann–Whitney U-tests were used to test differences
in the functional (i.e. “ecological guild”) composition of fish
assemblages between the two methods (see Table 1 and results
for metrics). Three guilds (spawning, trophic and habitat) with
simple and robust guild variables were considered, beside the
number/proportion of non-native species, which is an important
variable in environmental assessment studies. The classification
of species to a given guild variable followed Goffaux et al. (2005).
Species which were lacking from this study were categorized using
the system of Balon (1981) for spawning guild and the simplified
system of Erős et al. (2008) for trophic and habitat guild variables. To
increase the robustness of the analysis, statistical evaluations were
made only for the most abundant guild categories (see Table 2).
As for NMDS, samples were pooled for each site and collection
method used, and consequently, comparisons for each metric were

Table 2
Median values of different fish assemblage variables (number and proportion of guild variables) based on the samples collected with gillnetting or electric fishing

Gillnet Electric fishing P

Number of non-native species 0 2 0.009

Rel. abundance (%) of non-native species 0.0 5.9 0.009
Rel. biomass (%) of non-native species 0.0 37.6 0.009
Number of omnivorous species 3 4 NS
Rel. abundance (%) of omnivorous species 7.7 12.8 NS
Rel. biomass (%) of omnivorous species 43.5 26.2 NS
Number of insectivorous/piscivorous and piscivorous species 2 3 NS
Rel. abundance (%) of insectivorous/piscivorous and piscivorous species 0.8 7.7 0.009
Rel. biomass (%) of insectivorous/piscivorous and piscivorous species 2.3 41.7 0.009
Number of phytolithophilic species 5 4 0.037
Rel. abundance (%) of phytolithophylic species 99.1 76.3 0.009
Rel. biomass (%) of phytolithophylic species 89.3 10.2 0.009
Number of phytophilic species 1 3 0.016
Rel. abundance (%) of phytophylic species 0.6 12.6 0.009
Rel. biomass (%) of phytophylic species 10.1 26.0 0.047
Number of benthic species 3 2 NS
Rel. abundance (%) of benthic species 2.1 1.8 NS
Rel. biomass (%) of benthic species 21.0 23.9 NS

Differences were tested with Mann–Whitney U-test; NS, non-significant.

 T. Erős et al. / Fisheries Research 96 (2009) 70–76 73

made based on five, five samples for electric fishing and gillnetting,
respectively.

3. Results

Of the 20 species caught, 11 species were common for both

methods, while 6 and 3 species were caught exclusively either by
electric fishing or gillnetting, respectively (Table 1). For both meth-
ods, there were no differences in the number of species caught in
35 m long sampling units (Fig. 1a), although electric fishing samples
proved to be much more species rich when the samples were stan-
dardized according to the number of individuals collected (Fig. 1b).
The NPUE ± SD was 250 ± 252 for multi-mesh gillnets and 162 ± 173
for electric fishing.
Ordination of fish assemblage composition by NMDS showed
clear differences between the sampling methods based on both
relative abundance and relative biomass data (Fig. 2). Analysis
of relative abundance data including bleak (Fig. 2a) indicated
the dominance of this species in gillnet catches compared with
electric fishing, where rudd Scardinius erythrophthalmus, “rare
species” (especially eel Anguilla anguilla, bitterling Rhodeus sericeus,
pikeperch Sander lucioperca, and pumpkinseed Lepomis gibbosus)
and roach were relatively more dominant. A similar analysis with-
out bleak (Fig. 2b) reflected the higher relative abundance of roach,
perch Perca fluviatilis and white bream Blicca bjoerkna in gillnet
catches, in contrast to electric fishing, where bitterling, eel, rudd
and some rare species (especially asp Aspius aspius, gibel carp
Carassius gibelio, pike Esox lucius) proved to be the most impor-
tant species responsible for the separation. Ordination of relative
biomass data (Fig. 2c) revealed the importance of further species
such as pike, gibel carp and carp Cyprinus carpio in the differences
in fish assemblage composition between the two methods.
Comparison of length-frequency distributions based on all
species combined (i.e. the total sample) showed significant dif-

Fig. 2. Ordination by nonmetric multidimensional scaling (NMDS) of five reed habi-

tats based on the composition of the fish assemblages collected with electric fishing
(open circles) and gillnetting (filled squares). Data are as follows: (a) relative abun-
dance data for the whole assemblage, (b) relative abundance data without bleak and
(c) relative biomass data. The species shown beside the plots were significantly cor-
related with the ordination axis (Spearman’s rho in parenthesis, *P < 0.05; **P < 0.01;
***P < 0.001). For the definition of “rare species” see Section 2.2.

ferences (Chi-square test) between electric fishing and gillnetting

Fig. 1. Estimated number of species (±S.D.) as a function of (a) number of sam-
ples and (b) number of individuals collected with electric fishing (open circles) and
gillnetting (filled squares).
(Fig. 3; P < 0.001). Electric fishing caught both small (i.e. <5 cm) and
large (i.e. >25 cm) individuals in greater proportions and specimens
larger than 50 cm were caught exclusively with this method. The
length-frequency distributions of the two most abundant fishes,
74 T. Erős et al. / Fisheries Research 96 (2009) 70–76
Fig. 3. Length class distributions of all fish combined for electric fishing (open bars)
and gillnetting (filled bars).
bleak and roach, showed significant differences between electric
fishing and gillnetting for bleak (P < 0.001), whereas the differences
were not significant (P > 0.1) for roach (Fig. 4). Since the gillnet was
found to be ineffective in catching fish <5 cm (Appelberg, 2000;
CEN, 2005), we performed an additional analysis by omitting all
fish below 5 cm. The comparison also yielded significant between-
method differences in length-frequency histograms (P < 0.001).
Significant differences (Mann–Whitney U-test) were found
between the two sampling methods in the number and proportion
of species belonging to various ecological guilds (Table 2).
4. Discussion
Our study illustrates that reliance on single-gear surveys can
be misleading in assessing fish assemblages in reed habitats of a
shallow, eutrophic lake. Both electric fishing and gillnetting could
Fig. 4. Length class distributions of (a) bleak and (b) roach collected with electric
fishing (open bars) and gillnetting (filled bars).
complement each other when examining species richness, rela-
tive abundance/biomass relations, length-frequency distributions
and the composition of ecological guilds. However, the relative
effectiveness of one method over the other depends largely on
(1) the examined attribute of the fish assemblage, (2) the spe-
cific selectivity of the electric fishing gear and gillnets used in this
study, (3) the species composition, abundance and size relations
of the assemblage, including the specific behaviour of individual
fish species, and (4) the abiotic conditions of reed habitats in Lake
Balaton.
Electric fishing yielded more species than gill netting, show-
ing similarity to other studies (Growns et al., 1996; Goffaux et al.,
2005). However, the effectiveness of the two methods in detect-
ing new species depended largely on sampling effort (i.e. number
of samples). In fact, gillnetting was slightly more effective than
electric fishing at low sample sizes, but the species detection
rate declined more sharply after 20 samples. On the contrary, the
sample-based rarefaction curve of electric fishing catches did not
give even approximate saturation, but showed a slight but contin-
uous increase with the number of samples, suggesting that with
more effort additional species could have been detected. Since
the species pool of the lake is well known from many decades of
research (e.g. Bíró, 1997; Specziár et al., 1997), we can estimate
that ca. five additional species could have been detected with more
intensive electric fishing, including those found exclusively in gill-
net catches. However, these additional species live in other habitat
types (e.g. the pelagic razor fish Pelecus cultratus) or are very rare
in the lake (e.g. mud loach Misgurnus fossilis) and therefore can
appear in the catches sporadically and/or at extremely high sam-
pling effort.
In the few other studies in which similar sampling methods were
used (Growns et al., 1996; Goffaux et al., 2005), much higher differ-
ences in species detection were found between electric fishing and
gillnetting (but see Vaux et al., 2000, who used a low-powered elec-
tric fishing gear). Individual-based rarefaction curves indicated that
the relatively low effectiveness of electric fishing in detecting new
species in the function of samples was possibly due to low catch-
ing effectiveness (162 ind h−1 ). Since gillnet catches proved that the
density of fish was high in reed habitats in the eutrophic Lake Bal-
aton (250 ind net−1 h−1 ), it is likely that low water transparency
can be the main reason for the relatively poor performance of elec-
tric fishing in this shallow and therefore frequently mixed, turbid
water. When changes in the number of species were modelled in
the function of the number of individuals sampled, electric fishing
highly outperformed gillnetting. Overall, these results on species
diversity patterns suggest that both methods can be relatively use-
ful for detecting number of species and presence–absence patterns
in the reed habitats of Lake Balaton but that combining the two
methods can be more advantageous. Electric fishing is more effec-
tive if staffing, time and sampling costs allow intensive sampling
for collecting a large amount of fish, but gillnet sampling can be as
effective as electric fishing if only a limited number of samples can
be collected.
The high standard deviation values of NPUE data indicated
remarkable variability for both sampling methods. The examined
fish assemblage comprised mainly cyprinids which form shoals of
various size; this phenomenon may partly explain such high vari-
ability. However, the partly separated open water patches in the
reed habitats could also have affected the segregation of fish, which,
in turn, may have magnified spatial heterogeneities in fish abun-
dance. In fact, we could clearly observe the patchy distribution of
fish when catching them by electric fishing (the active method),
although the environmental variables across sections were appar-
ently the same. These results call attention to the importance of
spatially intensive surveys when assessing the fish assemblages of
 T. Erős et al. / Fisheries Research 96 (2009) 70–76
the littoral zones of lakes with high habitat complexity, such as reed
habitats.
Between-method differences based on relative abundance and
biomass data confirmed the results of species detection analysis
and showed that individual fish could be collected with relatively
low efficiency with electric fishing. Although bleak was the most
abundant fish in both catches, big shoals (i.e. >30 ind) could be only
occasionally collected with this method. On the contrary, bleak did
not avoid the net and was found in high numbers in gillnet catches.
Nevertheless, electric fishing was relatively more efficient in catch-
ing surface or water column fish (rudd, roach) compared with
benthic species (white bream) that were also abundant. This further
shaped differences between the two gears. These data also prove
that water transparency, and additionally water depth, affected the
catching effectiveness of electric fishing. The differences were also
reinforced by the stronger species and/or size selectivity of the
gillnet. Eel and bitterling were the most important species in this
respect, supporting the study of Goffaux et al. (2005) on riverine
fish assemblages. In addition, relative biomass data revealed the
constraints of multi-mesh gillnets in catching relatively big (e.g.
carp) and/or sedentary fish (e.g. the ambush predator pike). The
length-frequency analysis of data supports these findings well.
Although the EU standard gillnet showed limitations in catching
fish both from the lower and upper length classes, the two meth-
ods showed relatively high consistency in sampling intermediate
length classes of common cyprinids, such as bleak and roach, even
if length-frequency histograms differed significantly between the
two sampling methods for the former species. Our data suggest
that electric fishing should be preferred if only a single method
can be used for sampling littoral zone reed habitats for fish length
data. It is hard to design multi-mesh gillnets for catching small fish
(<5 cm) effectively, because smaller individuals usually move less
and when encountering the net are caught less effectively due to
their slower speed and the lower flexibility of small mesh sizes
(Olin and Malinen, 2003). However, the low efficiency of the EU
standard gillnet in catching large (>50 cm) fish can be improved by
using complementary gillnets of larger mesh sizes (e.g. 65, 80, 100
and 120 mm).
Although significant between-method differences were found
in the number of species in many aspects of fish assemblage catego-
rization (i.e. number of native vs. non-native species, components
of trophic, spawning and habitat guilds), these results should be
considered with caution, due to the difficulties in standardizing the
two methods for species richness comparisons. Evaluation of rela-
tive abundance and relative biomass data also showed substantial
between-method differences in the ratio of different groups. The
examined variables (e.g. relative abundance/biomass of non-native,
piscivorous or phytophilic species) are important in obtaining infer-
ences about biotic integrity or ecosystem functioning. Their reliable
estimation is therefore essential for responsible lake management.
However, it is hard to tell which method gave the better estimation
since values for the real community are unknown.
Studies, which use single-gear surveys to compare the fish
assemblages of different aquatic habitats, may do that with high
precision in the statistical sense (see e.g. Zar, 1999) if sampling
effort is high. These results can therefore be used for relative spa-
tial or temporal comparisons to a certain extent (e.g. Jeppesen et al.,
2006; Erős et al., 2008). However, since the accuracy of the samples
is unknown, the absolute bias of these estimates (i.e. their deviation
from the actual values which could be measured if the real/whole
assemblage could be sampled) is also unknown, as is the case in
our study. Inferences about the examined assemblages/system(s)
can be misleading if sampling accuracy is low, even if sampling
precision is high (not to mention if both accuracy and precision
are low). Although high sampling effort and the use of differ-
75
ent gear types may give reason for more reliable interpretations
(Jackson and Harvey, 1997) the real structure of the fish assemblage
may remain unknown for large habitats with diverse communities.
Nevertheless, increasing between-gear consistency with increasing
sampling effort may improve confidence in the data. Our data sug-
gest that, although costly for fish-based evaluations of ecosystem
health, monitoring should be based on multiple gears. The results
should then be evaluated separately and/or in the form of a com-
posite index to make more reliable comparisons of fish assemblages
within and between habitats.
Acknowledgements
We would like to express our thanks to Géza Dobos and András
Sevcsik for their help in the field work. This research was par-
tially funded by the Hungarian Scientific Research Fund (OTKA No.
T046222).
References
Appelberg, M., 2000. Swedish standard methods for sampling freshwater fish with
multi-mesh gillnets. Fiskerieverket Information 1, 3–32.
Appelberg, M., Berger, H.M., Hesthagen, T., Kleiven, E., Kurkilahti, M., Raitaniemi, J.,
Rask, M., 1995. Development and intercalibration of methods in Nordic fresh-
water fish monitoring. Water Air Soil Pollut. 85, 401–406.
Balon, E.K., 1981. Additions and amendments to the classification of reproductive
styles in fishes. Environ. Biol. Fish 6, 377–389.
Bíró, P., 1997. Temporal variations in Lake Balaton and its fish populations. Ecol.
Freshwater Fish. 6, 196–216.
Cao, Y., Hawkins, C.P., Vinson, M.R., 2003. Measuring and controlling data qual-
ity in biological assemblage surveys with special reference to stream benthic
macroinvertebrates. Freshwater Biol. 48, 1898–1911.
Casselman, J.M., Penczak, T., Carl, L., Mann, R.H.K., Holčík, J., Woitowich, W.A., 1990.
An evaluation of fish sampling methodologies for large river systems. Pol. Arch.
Hydrobiol. 37, 521–551.
CEN (European Committee for Standardization), 2003. Water quality—sampling of
fish with electricity. Brussels, 18 p.
CEN (European Committee for Standardization), 2005. Water quality—sampling of
fish with multi-mesh gillnets. Brussels, 27 p.
Colwell, R.K., 2005. EstimateS: statistical estimation of species richness and shared
species from samples. Version 7.5. Persistent URL <purl.oclc.org./estimates>.
Cowx, I.G. (Ed.), 1996. Fishing News Books. Blackwell Science, Oxford.
Cowx, I.G., Nunn, A.D., Harvey, J.P., 2001. Quantitative sampling of 0-group fish pop-
ulations in large lowland rivers: point abundance sampling by electric fishing
versus micromesh seine netting. Arch. Hydrobiol. 151, 369–382.
Diekmann, M., Brämick, U., Lemcke, R., Mehner, T., 2005. Habitat-specific fishing
revealed distinct indicator species in German lowland lake fish communities. J.
Appl. Ecol. 42, 901–909.
Drake, M.T., Valley, R.D., 2005. Validation and application of a fish-based index of
biotic integrity for small central Minnesota lakes. N. Am. J. Fish. Manage. 25,
1095–1111.
Erős, T., Tóth, B., Sevcsik, A., Schmera, D., 2008. Comparison of fish assemblage diver-
sity in natural and artificial rip-rap habitats in the littoral zone of a large river
(River Danube, Hungary). Int. Rev. Hydrobiol. 93, 88–105.
EU 2000. Directive 2000/60/EC of the European Parliament 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.
Garcia, X.F., Diekmann, M., Brämick, U., Lemcke, R., Mehner, T., 2006. Correlations
between type-indicator fish species and lake productivity in German lowland
lakes. J. Fish Biol. 68, 1144–1157.
Gassner, H., Tischler, G., Wanzenböck, J., 2003. Ecological integrity assessment of
lakes using fish communities—suggestions of new metrics developed in two
Austrian prealpine lakes. Int. Rev. Hydrobiol. 88, 635–652.
Goffaux, D., Grenouillett, G., Kestemont, P., 2005. Electric fishing versus gillnet sam-
pling for the assessment of fish assemblages in large rivers. Arch. Hydrobiol. 162,
73–90.
Gotelli, N.J., Colwell, R.K., 2001. Quantifying biodiversity: procedures and pitfalls in
the measurement and comparison of species richness. Ecol. Lett. 4, 379–391.
Growns, I.O., Pollard, D.A., Harris, J.H., 1996. A comparison of electric fishing and
gillnetting to examine the effects of anthropogenic disturbance on riverine fish
communities. Fish. Man. Ecol. 3, 13–24.
Istvánovics, V., Clement, A., Somlyódi, L., Specziár, A., Tóth, L.G., Padisák, J., 2007.
Updating water quality targets for shallow Lake Balaton (Hungary), recovering
from eutrophication. Hydrobiologia 581, 305–318.
Jackson, D.A., Harvey, H.H., 1997. Qualitative and quantitative sampling of lake fish
communities. Can. J. Fish. Aquat. Sci. 54, 2807–2813.
Jeppesen, E., Pekcan-Hekim, Z., Lauridsen, T.L., Søndergaard, M., Jensen, J.P., 2006.
Habitat distribution of fish in late summer: changes along a nutrient gradient in
Danish lakes. Ecol. Freshwater Fish. 15, 180–190.
76 T. Erős et al. / Fisheries Research 96 (2009) 70–76
Kennard, M.J., Pusey, B.J., Harch, B.D., Dore, E., Arthington, A.H., 2006. Estimating
local stream fish assemblage attributes: sampling effort and efficiency at two
spatial scales. Mar. Freshwater Res. 57, 635–653.
McCune, B., Mefford, M. J., 1997. PC-ORD. Multivariate analysis of ecological data.
Version 3.0. Glenede Beach, Oregon, MjM Software Design.
Mehner, T., Diekmann, M., Brämick, U., Lemcke, R., 2005. Composition of
fish communities in German lakes as related to lake morphology, trophic
state, shore structure and human-use intensity. Freshwater Biol. 50,
70–85.
Murphy, B.R., Willis, D.W. (Eds.), 1996. Fisheries Techniques, second ed. American
Fisheries Society, Bethesda, Maryland.
Olin, M., Malinen, T., 2003. Comparison of gillnet and trawl in diurnal fish community
sampling. Hydrobiologia 506–509, 443–449.
Olin, M., Kurkilahti, M., Peitola, P., Ruuhijärvi, J., 2004. The effects of fish accumulation
on the catchability of multimesh gillnet. Fish. Res. 68, 135–147.
Randall, R.G., Minns, C.K., 2002. Comparison of a habitat productivity index (HPI)
and an index of biotic integrity (IBI) for measuring the productive capacity of
habitat in nearshore areas of the Great Lakes. J. Great Lakes Res. 28, 240–255.
Specziár, A., 2001. Impacts of the activity of fish on the results of gillnet samplings:
diurnal and seasonal changes of the CPUE in Lake Balaton. Hidrol. Közl. 81 (5/6),
459–461 (in Hungarian with a summary in English).
Specziár, A., Vörös, L., 2001. Long-term dynamics of Lake Balaton’s chironomid fauna
and its dependence on the phytoplankton production. Arch. Hydrobiol. 152,
119–142.
Specziár, A., Tölg, L., Bíró, P., 1997. The fish stock structure in the reed stands of Lake
Balaton. Állattan. Közl. 82, 109–116 (in Hungarian with a summary in English).
Vaux, P.D., Whittier, T.R., DeCesare, G., Kurtenbach, J.P., 2000. Evaluation of a back-
pack electric fishing unit for multiple lake surveys of fish assemblage structure.
N. Am. J. Fish. Manage. 20, 168–179.
Zar, J.H., 1999. Biostatistical Analysis, fourth ed. Prentice Hall, New Jersey.