Openness of Fish Habitat Matters: Lake Pelagic Fish Community Starts Very Close to the Shore

Fish communities differ significantly between the littoral and the pelagic habitats. This paper attempts to define the shift in communities between the two habitats based on the European standard gillnet catch. We sampled the benthic and pelagic habitats from shore to shore in Lake Most and Římov Reservoir (Czech Republic). The 3 m deep pelagic nets were spanned across the water body at equal distances from two boundary points, where the depth was 3.5 m. The benthic community contained more fish, more species, and smaller individuals. The mild sloped littoral with a soft bottom attracted more fish than the sloping bank with a hard bottom and less benthos and large Daphnia. The catch of the pelagic nets was dominated by eurytopic fish—rudd (Scardinius erythrophthalmus) and roach (Rutilus rutilus) in Most and bleak (Alburnus alburnus) in Římov. With the exception of one case where overgrown macrophytes extended the structured habitat, the largest shift from the benthic to the pelagic community was observed only in the first pelagic gillnet above the bottom depth of 3.5 m. Open water catches were relatively constant with small signs of decline towards the middle of the lake. The results indicate that the benthic gillnet catch is representative of a very limited area and volume, while most of the volume is dominated by the pelagic community. This has important consequences for the assessment of the community parameters of the whole lake following the European standards for gillnet sampling design.


Introduction
In any environment, species composition changes gradually or abruptly between habitats. These ecological gradients have been the subject of a number of studies in ecology and usually reflect the abundance and richness of species [1][2][3][4]. Rapid changes in species ecological gradients, termed ecotones, have been observed in a variety of ecosystems [1]. Ecotones can affect the abundance and distribution of organisms.

Sampling Sites
Two water bodies in the Czech Republic, Lake Most and theŘímov Reservoir, were chosen for the experiment (Figure 1). Most is a post-coal mining lake (Ústí nad Labem region, 50.54 N, 13.65 E, see Figure 1) with an area of 310 ha, a maximum depth of 75 m, and a mean depth of 22 m. The lake was formed after the termination of coal mining in the summer of 1999 due to the filling of the open pit from autumn 2008 to autumn 2012. Most is an oligotrophic lake with a high abundance of macrophytes in its littoral area and a high water transparency [37]. paracol Water 2021, 13, x FOR PEER REVIEW 3 of 20 Two water bodies in the Czech Republic, Lake Most and the Římov Reservoir, were chosen for the experiment (Figure 1). Most is a post-coal mining lake (Ú stí nad Labem region, 50.54 N, 13.65 E, see Figure 1) with an area of 310 ha, a maximum depth of 75 m, and a mean depth of 22 m. The lake was formed after the termination of coal mining in the summer of 1999 due to the filling of the open pit from autumn 2008 to autumn 2012. Most is an oligotrophic lake with a high abundance of macrophytes in its littoral area and a high water transparency [37]. The Římov Reservoir (South Bohemia Region, 48.848 N, 14.487 E) is a canyon-shaped reservoir with a narrow (max. width 600 m) and elongated shape (length 10 km). The reservoir was built during the 1970s (from 1971 to 1978) and covers an area of about 200 ha with a volume of 34.3 × 10 6 m 3 , a maximum depth of 40 m, and an average depth of 12 m. Compared to the Most Lake, the Římov Reservoir has a gently to steeply sloping shore without submerged vegetation, which is missing due to significant water level fluctuations and low water transparency due to the eutrophic status of the water [38].

Gillnet Sampling in General
The European Standard gillnets (ESG) [39] were used to estimate the association of fish with littoral and pelagic habitats. The benthic ESG gillnet with 1.5 m height × 30 m length and 2.5 m mesh panels for each of the 12 mesh sizes was deployed in the littoral, while the pelagic gillnet with 3 m height × 30 m length and 2.5 m mesh panels for each of the 12 mesh sizes was deployed in the open water. The mesh sizes of the ESG followed a geometric series with a ratio of approximately 1.25 (5, 6.25, 8, 10, 12.5, 15.5, 19.5, 24, 29, 35, 43, and 55 mm) in random order. The first pelagic gill net from the shore was deployed above the bottom depth of 3.5 m (bottom line of the gillnet 0.5 m above the bottom). Depth was measured using a Humminbird Piranha echo sounder operating at 200 kHz.
Gillnet deployment was from bank to bank (Figures 2 and 3). The opposite banks differed in bottom slope, as is common in riverine waterbodies, and the fish community is influenced by the slope [40]. The mild sites had a bank slope of less than 8°, while the steep sites had a slope of more than 15°. The pelagic nets were laid out equidistantly, from the first pelagic net of the mild side to the first pelagic net of the steep side (Figures 2 and  3). The gillnets were named according to the slope of the bank on which they were deployed. The benthic gillnets were named MB (mild benthic-benthic net on the mild slope) and SB (steep benthic), while the pelagic gillnets were named MP (mild pelagic- TheŘímov Reservoir (South Bohemia Region, 48.848 N, 14.487 E) is a canyon-shaped reservoir with a narrow (max. width 600 m) and elongated shape (length 10 km). The reservoir was built during the 1970s (from 1971 to 1978) and covers an area of about 200 ha with a volume of 34.3 × 10 6 m 3 , a maximum depth of 40 m, and an average depth of 12 m. Compared to the Most Lake, theŘímov Reservoir has a gently to steeply sloping shore without submerged vegetation, which is missing due to significant water level fluctuations and low water transparency due to the eutrophic status of the water [38].

Gillnet Sampling in General
The European Standard gillnets (ESG) [39] were used to estimate the association of fish with littoral and pelagic habitats. The benthic ESG gillnet with 1.5 m height × 30 m length and 2.5 m mesh panels for each of the 12 mesh sizes was deployed in the littoral, while the pelagic gillnet with 3 m height × 30 m length and 2.5 m mesh panels for each of the 12 mesh sizes was deployed in the open water. The mesh sizes of the ESG followed a geometric series with a ratio of approximately 1.25 (5, 6.25, 8, 10, 12.5, 15.5, 19.5, 24, 29, 35, 43, and 55 mm) in random order. The first pelagic gill net from the shore was deployed above the bottom depth of 3.5 m (bottom line of the gillnet 0.5 m above the bottom). Depth was measured using a Humminbird Piranha echo sounder operating at 200 kHz.
Gillnet deployment was from bank to bank (Figures 2 and 3). The opposite banks differed in bottom slope, as is common in riverine waterbodies, and the fish community is influenced by the slope [40]. The mild sites had a bank slope of less than 8 • , while the steep sites had a slope of more than 15 • . The pelagic nets were laid out equidistantly, from the first pelagic net of the mild side to the first pelagic net of the steep side (Figures 2 and 3). The gillnets were named according to the slope of the bank on which they were deployed.
The benthic gillnets were named MB (mild benthic-benthic net on the mild slope) and SB (steep benthic), while the pelagic gillnets were named MP (mild pelagic-pelagic net on the side, adjacent to the mild slope) and SP (steep pelagic), with one pelagic gillnet deployed in the center of the lake (mid distance between the two 3.5 m isobaths in the sampled area) referred to as the center net. Given that the pelagic area had more nets, the number immediately following the acronym indicates the number of gillnets deployed from the shore, e.g., SP1 is the first pelagic gillnet from the steep shore, while SP2 is the following pelagic gillnet, and so on (Figures 2 and 3).
Water 2021, 13, x FOR PEER REVIEW 4 of 20 pelagic net on the side, adjacent to the mild slope) and SP (steep pelagic), with one pelagic gillnet deployed in the center of the lake (mid distance between the two 3.5 m isobaths in the sampled area) referred to as the center net. Given that the pelagic area had more nets, the number immediately following the acronym indicates the number of gillnets deployed from the shore, e.g., SP1 is the first pelagic gillnet from the steep shore, while SP2 is the following pelagic gillnet, and so on (Figures 2 and 3).  Sampling was done in accordance with CEN standards [39], with the gillnets being deployed 2 h before the sunset and lifted 2 h after the sunrise [41]. Standard fish length and weight of all the captured individuals were measured to the nearest mm and g, respectively. The catch per unit of effort (CPUE) was defined as the number of individuals per 1000 m 2 of net per night, analogically the sampled biomass per unit of effort (BPUE) was defined as number of grams per 1000 m 2 of a net per night.  pelagic net on the side, adjacent to the mild slope) and SP (steep pelagic), with one pelagic gillnet deployed in the center of the lake (mid distance between the two 3.5 m isobaths in the sampled area) referred to as the center net. Given that the pelagic area had more nets, the number immediately following the acronym indicates the number of gillnets deployed from the shore, e.g., SP1 is the first pelagic gillnet from the steep shore, while SP2 is the following pelagic gillnet, and so on (Figures 2 and 3).  Sampling was done in accordance with CEN standards [39], with the gillnets being deployed 2 h before the sunset and lifted 2 h after the sunrise [41]. Standard fish length and weight of all the captured individuals were measured to the nearest mm and g, respectively. The catch per unit of effort (CPUE) was defined as the number of individuals per 1000 m 2 of net per night, analogically the sampled biomass per unit of effort (BPUE) was defined as number of grams per 1000 m 2 of a net per night. Sampling was done in accordance with CEN standards [39], with the gillnets being deployed 2 h before the sunset and lifted 2 h after the sunrise [41]. Standard fish length and weight of all the captured individuals were measured to the nearest mm and g, respectively. The catch per unit of effort (CPUE) was defined as the number of individuals per 1000 m 2 of net per night, analogically the sampled biomass per unit of effort (BPUE) was defined as number of grams per 1000 m 2 of a net per night.

Most Lake Sampling Design
Sampling in Lake Most was conducted from 3-6 September 2018. The transect area in Lake Most was in the form of a ribbon that extended from shore to shore across the lake. We sampled two benthic and nine pelagic sites in the lake ( Figure 2). The east mild shore had a slope of 7 • of declination and the west steep shore had a slope of 15 • . At each littoral or pelagic location shown in Figure 2, we set up three ESG nets connected by a 40 m rope to ensure adequate spacing between them. The distance between gillnet sampling locations was 150 m. The benthic gillnets were deployed at a depth of 1.5-2.5 m, and the first pelagic gillnet was deployed at a depth greater than 3.5 m from each bank. The remaining pelagic gillnets were deployed at the same spacing. The gillnets were deployed parallel to the shore. Altogether 6 benthic and 27 pelagic gillnets were deployed.

2.4.Římov Reservoir Sampling Design
Sampling of theŘímov Reservoir was conducted from 30 July to 2 August 2019. Six locations in the reservoir were sampled, each with both a mild slope shore (2 • to 8 • slope) and a steeply sloping bank (20 • to 35 • slope). We selected sites only in the true lacustrine zone ( Figure 3) to avoid the change in productivity that increases further upstream [42]. A single ESG device was deployed at each net location of each site. The nets were scattered to ensure that no net interfered with the others ( Figure 3B). The minimum distance between nets was 60 m, but usually it was more than 100 m. For this experiment, we also deployed two sets of benthic gillnets on either side of the reservoir (one in the 0-1.5 m depth range and the second in 1.5-3 m). For this article, the CPUE and BPUE values from these two nets were combined so that they well represent the littoral range of 0-3 m. Altogether, 24 benthic and 42 pelagic gillnets were deployed.
Zooplankton samples were collected 30-60 min after each gillnet deployment. Vertical hauls with a plankton net (diameter 20 cm, mesh size 0.2 mm) were made at both ends of each pelagic net. Hauls were made from 3 m depth to the surface and two hauls were combined in each zooplankton sample. Samples of littoral zooplankton were collected using a Schindler sampler (volume 30 L, mesh size 0.2 mm). Each sample of littoral zooplankton was collected by combining two samples (one from the upper, 0-1.5 m, and one from the lower, 1.5-3 m, portion of the sampled layer up to 3 m) in one bottle. Samples of littoral zooplankton were collected from both ends of the benthic gillnets deployed in the littoral zones. The zooplankton was divided into 3 groups: Daphnia galeata, other Cladocera (Acroperus harpae, Bosmina coregoni, Bosmina longirostris, Ceriodaphnia quadrangula, Diaphanosoma brachyurum, Chydorus sphaericus, Leptodora kindti, and Leydigia leydigi), and Copepoda (Cyclops vicinus, Eudiaptomus gracilis, Mesocyclops leuckarti, Thermocyclops crassus, Thermocyclops oithonoides, Cyclopoida-copepodites, and Diaptomidae-copepodites). In addition, 100 individuals of D. galeata were measured for body size in each zooplankton sample. The body size of the Daphnia was measured from the top of the head to the base of the caudal spine. An amount of 1 mm of the body length of the Daphnia was chosen as the threshold between the small and the large individuals.

Data Analyses
Catch per unit effort (CPUE) was calculated as the mean of the total number of individuals divided by the total sampling effort (net surface area), while biomass per unit effort (BPUE) was calculated as the total weight of catch per 1000 m 2 of net area. CPUE and BPUE were calculated for individual species as well as for the entire fish assemblage.
Negative binomial generalized linear models (GLM) were applied to describe the differences in fish CPUE and BPUE values (CPUE and BPUE) with distance from shore in Most. The negative binomial generalized linear model was chosen because it can cope with a large number of zeros and over-dispersed data [44]. The MASS package was used to compute all GLMs [45].
For the analyses in theŘímov Reservoir, a generalized linear mixed effects model (GLMM) fitted for the negative binomial family was used, with localities included in the model as a random effect. The model was applied to describe differences in fish CPUE and BPUE and zooplankton density as a function of distance from shore, as well as benthic macroinvertebrate numbers on gentle and steep slopes and at different depths. All data analyses were performed using R software [46].
Diversity indices (Shannon-Weaver, Simpson, Pielou's evenness, and richness) of fish communities were also calculated using the Vegan package of the R software [47].
CPUE values for the entire fish community decreased sharply from the shore to the center of the lake, on both mild (p < 0.001, deviance = 18.6) and steep (p < 0.001, deviance = 15.5) shores ( Figure 4). This pattern is particularly striking from the first pelagic gill net SP1 to the middle gill net, which had the lowest CPUE values among all gill nets used (Table 1).
Water 2021, 13, x FOR PEER REVIEW 6 of 20

Data Analyses
Catch per unit effort (CPUE) was calculated as the mean of the total number of individuals divided by the total sampling effort (net surface area), while biomass per unit effort (BPUE) was calculated as the total weight of catch per 1000 m 2 of net area. CPUE and BPUE were calculated for individual species as well as for the entire fish assemblage.
Negative binomial generalized linear models (GLM) were applied to describe the differences in fish CPUE and BPUE values (CPUE and BPUE) with distance from shore in Most. The negative binomial generalized linear model was chosen because it can cope with a large number of zeros and over-dispersed data [44]. The MASS package was used to compute all GLMs [45].
For the analyses in the Římov Reservoir, a generalized linear mixed effects model (GLMM) fitted for the negative binomial family was used, with localities included in the model as a random effect. The model was applied to describe differences in fish CPUE and BPUE and zooplankton density as a function of distance from shore, as well as benthic macroinvertebrate numbers on gentle and steep slopes and at different depths. All data analyses were performed using R software [46].
Diversity indices (Shannon-Weaver, Simpson, Pielou's evenness, and richness) of fish communities were also calculated using the Vegan package of the R software [47].
CPUE values for the entire fish community decreased sharply from the shore to the center of the lake, on both mild (p < 0.001, deviance = 18.6) and steep (p < 0.001, deviance = 15.5) shores ( Figure 4). This pattern is particularly striking from the first pelagic gill net SP1 to the middle gill net, which had the lowest CPUE values among all gill nets used (Table 1).

Figure 4.
Total catch per unit effort (CPUE; individuals per 1000 m 2 of net) from 11 gillnets fished in Most Lake. The boxplot represents the quartile value of the CPUE, the grey dots represent the CPUE of three individual nets deployed at the same distance from shore, the thick middle line represents the median, and the white dot represents the arithmetic mean. The site MP1 was surrounded by overgrown macrophytes. Table 1. Catch per unit of effort (inds. 1000 m −2 of gillnets), standard errors (se), and the significance level (p) at various benthic and pelagic sites of Most Lake. See Section 2 for detailed explanations of gillnet locations. . Total catch per unit effort (CPUE; individuals per 1000 m 2 of net) from 11 gillnets fished in Most Lake. The boxplot represents the quartile value of the CPUE, the grey dots represent the CPUE of three individual nets deployed at the same distance from shore, the thick middle line represents the median, and the white dot represents the arithmetic mean. The site MP1 was surrounded by overgrown macrophytes. Table 1. Catch per unit of effort (inds. 1000 m −2 of gillnets), standard errors (se), and the significance level (p) at various benthic and pelagic sites of Most Lake. See Section 2 for detailed explanations of gillnet locations.  At the site of MP1 (bottom depth = 3.5 m), many high macrophyte stands were still present, so the habitat cannot be considered truly pelagic. This may have been the cause of the higher CPUE values at this site ( Figure 4). When analyzing each species independently, two of the five species showed significant non-random distribution from bank to bank, namely roach (mild: p < 0.001, deviance = 21.9; steep: p < 0.001, deviance = 18.3) and rudd (mild: p = 0.003, deviance = 21.5).
The BPUE values differed significantly for both the mild slope (p < 0.001, deviance = 18.9) and the steep shores (p < 0.023, deviance = 16, Figure 5, Table 2). The influence of the macrophyte beds at the site of MP1 was again very evident, with BPUE more than twice that of the other gillnets (Table 2). When analyzing the distribution of the individual species on both slopes, the results followed a similar pattern to the CPUE in the case of roach (mild: p < 0.001, deviance = 21.1; steep: p < 0.002, deviance = 18.2), with a significant response to distance from both shores, and also perch, but only for the mild slope side (mild: p < 0.001, deviance = 12.6).   At the site of MP1 (bottom depth = 3.5 m), many high macrophyte stands were still present, so the habitat cannot be considered truly pelagic. This may have been the cause of the higher CPUE values at this site ( Figure 4). When analyzing each species independently, two of the five species showed significant non-random distribution from bank to bank, namely roach (mild: p < 0.001, deviance = 21.9; steep: p < 0.001, deviance = 18.3) and rudd (mild: p = 0.003, deviance = 21.5).
The BPUE values differed significantly for both the mild slope (p < 0.001, deviance = 18.9) and the steep shores (p < 0.023, deviance = 16, Figure 5, Table 2). The influence of the macrophyte beds at the site of MP1 was again very evident, with BPUE more than twice that of the other gillnets (Table 2). When analyzing the distribution of the individual species on both slopes, the results followed a similar pattern to the CPUE in the case of roach (mild: p < 0.001, deviance = 21.1; steep: p < 0.002, deviance = 18.2), with a significant response to distance from both shores, and also perch, but only for the mild slope side (mild: p < 0.001, deviance = 12.6).      Rudd and roach clearly dominated the fish community of Most Lake ( Figure 6). They had absolute dominance in the open water habitats, while inshore at the benthic habitat, the dominance was shared with perch (and rarely with ruffe). The dominance of rudd was even more evident in the biomass (Figure 7). This means that, on average, rudd were larger than roach in open waters (see also Table 3). Analysis of the overall size distribution also showed that the lowest mean sizes were found in the first nets on each shore, and the highest mean size would be in the center (p < 0.001, deviance = 903.44). Rarely, larger individuals of pike and perch were also caught in the pelagic area (Figure 7). Perch dominated the benthic habitats in terms of biomass, which was significantly different from all pelagic habitats. The diversity indices of the fish community of Lake Most were generally low and showed a weak tendency to decrease towards the center of the lake (Figure 8). The low values correspond to a low number of species present. The presence of littoral elements such as perch and ruffe and the lower dominance of rudd resulted in a slightly higher diversity in the littoral. However, none of the diversity indices showed a significant trend between the littoral and the pelagic. Esox lucius  Rudd and roach clearly dominated the fish community of Most Lake ( Figure 6). They had absolute dominance in the open water habitats, while inshore at the benthic habitat, the dominance was shared with perch (and rarely with ruffe). The dominance of rudd was even more evident in the biomass (Figure 7). This means that, on average, rudd were larger than roach in open waters (see also Table 3). Analysis of the overall size distribution also showed that the lowest mean sizes were found in the first nets on each shore, and the highest mean size would be in the center (p < 0.001, deviance = 903.44). Rarely, larger individuals of pike and perch were also caught in the pelagic area (Figure 7). Perch dominated the benthic habitats in terms of biomass, which was significantly different from all pelagic habitats. The diversity indices of the fish community of Lake Most were generally low and showed a weak tendency to decrease towards the center of the lake (Figure 8). The low values correspond to a low number of species present. The presence of littoral elements such as perch and ruffe and the lower dominance of rudd resulted in a slightly higher diversity in the littoral. However, none of the diversity indices showed a significant trend between the littoral and the pelagic.
Bleak superdominance was evident in all the pelagic samples ( Figure 11). Only in the benthic samples on both sides of the lake did other species make up a larger proportion. When the biomass was expressed, the dominance of bleak persisted but was less evident (Figure 12). The species composition shows a gradual change from the shore to the open water, where the first pelagic net showed a species composition between the benthic and the pelagic habitat (still a conspicuous presence of roach, perch, bream, and asp). Ruffe is the best indicator of the benthic habitat, followed by the perch. Pikeperch, asp, catfish, and bream were caught in the open water, but their proportion was often lower than near shore. Rudd was not abundant, but also behaved like a eurytopic species, showing a homogeneous horizontal distribution. The distinct pattern of species distribution is reflected in a clear pattern of diversity indices ( Figure 13). Species richness and diversity were always highest in the nearshore habitat and decreased towards the center of the reservoir. The Shannon index (p = 0.0316) and the number of species (p = 0.0040) showed a significant negative trend from the littoral to the pelagic. The size distribution also showed that the lowest mean lengths were found in the first nets on each side of the reservoir and the highest mean length was found in the center (p < 0.001, deviance = 56,291.8, variance = 0.0004, Table 6).
The mean densities of Dapnia galeata (the main food of non-predatory fish) were slightly higher at the littoral of the mild slope but were not significantly different from the other sampling stations along the transverse profile, with the exception of sites SP2 and SP1 ( Figure 14). We also divided D. galeata into two size classes (small: body size ≤ 1 mm; large: body size > 1 mm) and tested whether the densities of these size classes differed along the cross-section. Densities of small and large Daphnia were higher on average at the littoral of the mild slope littoral but were generally not significantly different from the other sampling stations, except for SP2 (small Daphnia), SP1, and SB (large Daphnia; Figure 15). The other two groups of zooplankton, other Cladocera, and Copepoda, were evenly distributed across the transverse profile ( Figure 14).
The mean densities of Dapnia galeata (the main food of non-predatory fish) were slightly higher at the littoral of the mild slope but were not significantly different from the other sampling stations along the transverse profile, with the exception of sites SP2 and SP1 ( Figure 14). We also divided D. galeata into two size classes (small: body size ≤ 1 mm; large: body size > 1 mm) and tested whether the densities of these size classes differed along the cross-section. Densities of small and large Daphnia were higher on average at the littoral of the mild slope littoral but were generally not significantly different from the other sampling stations, except for SP2 (small Daphnia), SP1, and SB (large Daphnia; Figure  15). The other two groups of zooplankton, other Cladocera, and Copepoda, were evenly distributed across the transverse profile ( Figure 14).

Figure 14.
Mean density of three zooplankton groups (Daphnia galeata, other Cladocera, and Copepoda) at different distances from the shore of Římov Reservoir. Different letters indicate significant differences (p < 0.05) in D. galeata density between different distances from shore to shore. The densities of other Cladocera and Copepoda did not differ across the transverse profile (p > 0.05). Letters a and b denominate significant differences in D. galeata densities. Figure 14. Mean density of three zooplankton groups (Daphnia galeata, other Cladocera, and Copepoda) at different distances from the shore ofŘímov Reservoir. Different letters indicate significant differences (p < 0.05) in D. galeata density between different distances from shore to shore. The densities of other Cladocera and Copepoda did not differ across the transverse profile (p > 0.05). Letters a and b denominate significant differences in D. galeata densities.
Water 2021, 13, x FOR PEER REVIEW 15 of 20 Figure 15. Mean density of small (≤1 mm) and large (>1 mm) Daphnia galeata at different distances from the shore of Římov Reservoir. Significant differences (p < 0.05) in the density of small D. galeata between different distances from shore to shore are indicated by different lowercase letters. Significant differences (p < 0.05) in the density of large D. galeata between different distances from shore to shore are indicated by different uppercase letters.
Benthic macroinvertebrates were generally more abundant on the gentle slopes ( Figure 16). A significant difference between the mild and steep sites was found for the Chironomidae (p = 0.02, deviance = 101.6, variance < 0.001), Ephemeroptera (p < 0.001, deviance = 161.7, variance = 0.060), and permanent fauna (p < 0.001, deviance = 116.7, variance < 0.001) groups. For the difference in depth, only the permanent fauna was significantly less abundant in deeper water (p < 0.001, deviance = 123.2, variance < 0.001). Mean density of small (≤1 mm) and large (>1 mm) Daphnia galeata at different distances from the shore ofŘímov Reservoir. Significant differences (p < 0.05) in the density of small D. galeata between different distances from shore to shore are indicated by different lowercase letters. Significant differences (p < 0.05) in the density of large D. galeata between different distances from shore to shore are indicated by different uppercase letters.

Figure 15.
Mean density of small (≤1 mm) and large (>1 mm) Daphnia galeata at different distances from the shore of Římov Reservoir. Significant differences (p < 0.05) in the density of small D. galeata between different distances from shore to shore are indicated by different lowercase letters. Significant differences (p < 0.05) in the density of large D. galeata between different distances from shore to shore are indicated by different uppercase letters.

Discussion
Our experiments have shown that the fish community changes very abruptly from the littoral to the pelagic in two different systems just near the benthic habitat. At the first pelagic point only 0.5 m above the bottom, the proportion of littoral species abruptly

Discussion
Our experiments have shown that the fish community changes very abruptly from the littoral to the pelagic in two different systems just near the benthic habitat. At the first pelagic point only 0.5 m above the bottom, the proportion of littoral species abruptly decreased. The pelagic habitat showed a homogeneous fish community composition, with a slight gradient corresponding to the distance from the littoral. This result supports previous assumptions that the definition of the benthic habitat only applies within a few meters of the bottom, and that the assumed height of the benthic habitat of 1.5 m above the bottom [31,36] may be accurate. The exception was the mild slope of Lake Most (site MP1), where the presence of abundant macrophytes created conditions that were very different from the pelagic habitat. The results also support previous assumptions that the pelagic habitat is the main volume even in medium-sized lakes, and that large volumes of open water must be considered if representative fish community values are to be obtained for the entire lake. Our results provide reassurance that the volume of the pelagic habitat is as large as estimated in previous studies [33,35] and that it is by far the most important habitat, even in relatively small waters [36].
The majority of species showed that they were benthic-bound, such as perch, ruffe, bream, pikeperch, asp, and roach (in some BPUE, also catfish). Bleak and rudd were determined to be typical eurytopic species. No exclusively pelagic species was found, which is consistent with the theory of Fernando and Holčík [24] about the scarcity of truly pelagic fish in young ecosystems. Consequently, the transition from the littoral to the pelagic community is mainly characterised by a sharp decline in the abundance and proportion of benthic species. This reflects the fact that the pelagic community is much simpler and less diverse, with fewer fish species willing to leave the safety of the littoral [24,26]. The results of this study showed that the fish community changes very quickly on the way from the bottom to the open water, and what we may call the pelagic community shapes most of the volume of lakes and reservoirs. This supports previous studies indicating that volume-weighted estimates provide much more realistic estimates for entire lakes than the global CEN CPUE [33].
The gradual decrease in fish abundance from the littoral to the pelagic zone in the middle of the lake was more evident in Most than inŘímov. One reason for this difference could be the higher complexity of the habitat in the littoral of Most, due to the lower steepness and the high macrophyte density in the littoral zone, or the higher steepness inŘímov. Littoral aquatic macrophytes are important components of habitat complexity and heterogeneity, as they dominate the nearshore zones of lakes and support diverse fish communities [48,49]. Macrophytes can influence fish habitat selection and ecological relationships such as predation and competition, which in turn affect the fish community structure. For example, predators may induce their prey to seek shelter in roots, leaves, and stems, which act as visual and physical barriers and provide protection from predators [27,50], while competition may induce fish individuals to seek new feeding grounds and reproduce [51]. Macrophyte habitats are considered nursery grounds for juvenile fish because they provide numerous sheltering opportunities, as smaller fish are more vulnerable to predators than larger fish [52,53]. The high macrophyte stems most likely caused very high fish catches at the first pelagic net at the mild slope of Most Lake.
However, even in the habitat without true aquatic macrophytes (Římov Reservoir), the CPUE, BPUE, and species diversity were mostly higher in the benthic habitats. This indicates that for many species at least the presence of some substrate is also important. The comparison between the benthic net catches shows that the mild slope "beach-type" habitats contained more fish than the steep slopes. Although the steep slopes may be more structured by rocks and tree remains [30,54], they are more open and clearly less safe for prey fish (see also [40]). The soft bottom substrate of mild slope shores is more favorable for benthic macroinvertebrates, and habitats with gentle slopes have also been found to have slightly higher densities of cladoceran D. galeata. According to previous studies carried out in theŘímov Reservoir, cladoceran D. galeata is the main prey of the dominant non-predatory fish species [18,55,56]. Therefore, the reason for fish staying in the mild slope littoral of theŘímov Reservoir could be both the protection from predators and feeding on D. galeata and the available benthic resources [30]. Other Cladocera (mostly represented by small species such as Diaphanosoma brachyurum and Eubosmina coregoni) and Copepoda were evenly distributed across the transverse profile of this reservoir and therefore did not appear to affect fish distribution. In general, the lowest average fish lengths were found in the littoral habitats, suggesting that juvenile fish feel more secure in the nearshore zone. This is in general agreement with the results of previous studies from other limnetic ecosystems [27]. While fish densities in the littoral mild slope habitats were considerably higher than in the open water, the CPUE and BPUE in the littoral habitats with the steep slopes were similar to those in the pelagic area. This may also be because in the steep slope habitats, the first pelagic net above the 3.5 m isobath was necessarily very close to the shore.
Our study only has a horizontal dimension. It deals with a layer of 0-3 m, which is normally the most populated by fish [8,11]. It was beyond our capabilities to extend the study to deeper habitats. However, the results from the 3 m depth are quite convincing, and we cannot expect the situation to change significantly in further layers. Fishes that require the substrate tend to stay close to it [57], while eurytopic fishes disperse without much regard to the benthic habitat.

Conclusions
Our experiment showed that the littoral zone was characterized by high numbers of fish, especially perch, and by the presence of smaller individuals. The catch of the pelagic nets was dominated by eurytopic fish-rudd and roach in Most and bleak inŘímov. With the exception of one case where abundant macrophytes extended the structured habitat, the largest shift from the benthic to the pelagic community was observed only at the first pelagic gillnet at a bottom depth of 3.5 m. Open water catches were relatively consistent with small signs of a gradient towards the middle of the lake. The results indicate that the benthic gillnet catch is representative of a very limited area and volume, while most of the volume is dominated by the pelagic community, the most important habitat even in relatively small waters. This has important consequences for the assessment of community parameters of the whole lake.